CN112017436B - Method and system for predicting urban traffic travel time - Google Patents

Abstract

The invention relates to a method and system for predicting traffic travel time in a city. The predicting method includes dividing a city to be measured into rectangular grids; building a normalized floating-point traffic flow based on the rectangular grid and historical traffic flow data Matrix; according to the normalized floating-point traffic flow matrix, train the urban traffic flow prediction network; simplify the trajectory point sequence of the vehicle driving path according to the rectangular grid, and obtain the gridded trajectory of the vehicle driving path; according to the normalized Convert the floating-point traffic flow matrix and gridded trajectory to determine the eigenvectors of each trajectory point in the gridded trajectory; train the urban traffic travel time prediction network based on the eigenvectors of each trajectory point; based on the urban traffic flow prediction The network and urban travel time prediction network can accurately determine the travel time required for the vehicle to be tested to complete the path to be tested according to the travel path of the vehicle to be tested, which can improve the prediction accuracy in complex scenarios.

Description

Prediction method and system of urban traffic travel time

technical field

The invention relates to the technical field of urban road condition information processing, in particular to a method and system for predicting the travel time of urban traffic in combination with real-time road condition information.

Background technique

The travel time prediction of urban traffic vehicles is the basis for applications such as route planning, vehicle navigation, intelligent carpooling, and intelligent taxi order sharing. In the case of delay, predict the time required for the vehicle to complete the route. When travelers can consider multiple alternative routes, accurate travel time prediction results can help choose the optimal route while avoiding congested sections, which is very important for optimizing urban management and improving the travel efficiency of citizens.

At present, the traditional urban traffic travel time prediction methods are mainly divided into two categories, namely parametric methods and non-parametric methods. Parametric methods rely on existing mathematical and statistical theories to make predictions, including linear regression models, autoregressive integrated moving average models, Kalman filtering, Bayesian dynamic linear models, hidden Markov models, and support vector machines. These methods have achieved varying degrees of success in short-distance travel time, but their accuracy is difficult to meet the application requirements for long-distance travel time prediction. Non-parametric methods use a data-driven approach to capture the underlying knowledge of the data. The non-parametric method does not need to make assumptions about the applicable data distribution of the model, and has a certain data adaptive ability. Nonparametric methods include k-nearest neighbor estimation, voting decision, Gaussian process, etc. However, the parameter method has a large amount of calculation, and it is difficult to meet the requirements of real-time performance indicators.

Traditional parametric methods and non-parametric methods are currently difficult to apply to practical scenarios, mainly for the following two reasons. First, the traditional method only uses the historical urban traffic travel time data, ignoring the impact of real-time road conditions on the urban traffic travel time prediction. When the user needs to predict the travel time of a certain route, the traditional method is only based on the traffic conditions on the route at the current time point, and it is difficult to dynamically fuse the real-time road condition information. However, when the user travels to a certain road section, the road condition of the road section has changed, for example, from the original congested state to the current unblocked state, or from the original unblocked state to the current congested state. This change in road conditions is especially noticeable during long-distance driving. Second, the nonlinear fitting ability of traditional methods is limited. In urban areas with short-distance local scope and obvious changes in traffic conditions, traditional methods can usually obtain better prediction results of intra-city traffic travel time. However, for long-distance and large-scale areas, traffic conditions exhibit highly nonlinear dynamic characteristics, and the spatial and temporal correlations are very complex. Traditional methods are difficult to capture these complex relationships, so they tend to produce large prediction errors of urban traffic travel time.

In recent years, deep learning has flourished, and deep learning methods represented by Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) have been widely cited in a large number of tasks related to machine learning. It has achieved great success in many practical applications such as image recognition, image semantic segmentation, visual object detection, video content understanding, machine translation, and autonomous driving. Deep learning has powerful nonlinear fitting capabilities. At the same time, due to the flexibility of its network structure design, deep learning is conducive to the fusion and processing of multiple information.

To sum up, traditional methods are still difficult to achieve ideal results in dealing with long-distance intra-city traffic travel time prediction tasks in complex scenarios.

SUMMARY OF THE INVENTION

In order to solve the above problems in the prior art, that is, in order to improve the accuracy of urban traffic travel time prediction, the purpose of the present invention is to provide a method and system for predicting urban traffic travel time.

In order to solve the above-mentioned technical problems, the present invention provides the following scheme:

A method for predicting traffic travel time in a city, the predicting method comprising:

Divide the city to be tested into a rectangular grid;

Based on the rectangular grid, construct a normalized floating-point traffic flow matrix according to historical traffic flow data;

According to the normalized floating-point traffic flow matrix, train the urban traffic flow prediction network;

Simplify the trajectory point sequence of the vehicle's driving path according to the rectangular grid, and obtain the gridded trajectory of the vehicle's driving path;

According to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the eigenvectors of each trajectory point in the gridded trajectory;

According to the feature vector of each trajectory point in the gridded trajectory, train the urban traffic travel time prediction network;

Based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the travel path of the vehicle to be tested, the travel time required for the vehicle to be tested to complete the path to be tested is determined.

Optionally, constructing a normalized floating-point traffic flow matrix according to historical traffic flow data based on the rectangular grid specifically includes:

The history n _d days are selected as the training period, and each day of the training period is divided into K _d time intervals with fixed time intervals; the total number of continuous time intervals that the training period is divided into is K: K=n _d ×K _d ;

表示在第t个时间区间内出现在矩形网格G的第i行第j列格子区域内的车辆个数；Counting the trajectory data of vehicles entering the networked system in the rectangular grid G during the training period, and each trajectory data records the longitude position, latitude position and time information of the vehicle;

Indicates the number of vehicles appearing in the grid area of the i-th row and the j-th column of the rectangular grid G in the t-th time interval;

Traverse all the time intervals in the training period, count the traffic flow of vehicles entering the networked system in the same time interval, and obtain K traffic flow matrices, denoted as X ₁ , X ₂ ,...,X _k ,... ,X _K ; X _k represents the traffic flow matrix in the kth time interval, k=1,2,....,K;

According to the following formula, normalized floating-point processing is performed on each traffic flow matrix to obtain the corresponding normalized floating-point traffic flow matrix:

为第t个时间区间内的归一化浮点交通流量矩阵Y_t的第i行第j列元素，x₀表示K个交通流量矩阵中所有元素的最大值。in,

is the i-th row and j-th column element of the normalized floating-point traffic flow matrix Y _t in the t-th time interval, and x ₀ represents the maximum value of all elements in the K traffic flow matrices.

Optionally, the training of an urban traffic flow prediction network according to the normalized floating-point traffic flow matrix specifically includes:

According to the normalized floating-point traffic flow matrix, a corresponding first training sample is constructed:

P _t =(A _t ; B _t ),

Among them, t represents the sequence number of the time interval, P _t represents the first training sample; A _t represents the sample feature of the first training sample P _t , which is a three-dimensional tensor with L rows, M columns, and J layers, starting from the t-th time interval as the end point. The normalized traffic flow _matrices obtained in the _{first J consecutive} time _intervals of the ];

B _t represents the sample label of the first training sample P _t , which is a three-dimensional tensor with L rows and M columns and Q layers. The normalized traffic obtained in Q consecutive time intervals starting from the t+1th time interval The flow matrix is stacked in chronological order: B _t =[Y _t+1 ,Y _t+2 ,…,Y _t+Q ];

Among them, Y _t-J+1 is the normalized floating-point traffic flow matrix obtained in the t+1th time interval, Y _t is the normalized floating-point traffic flow matrix obtained in the t-th time interval, and Y _{t +Q} is the normalized floating-point traffic flow matrix obtained in the t+Qth time interval;

The urban traffic flow prediction network is trained according to the first training sample; the urban urban traffic flow prediction network is used to predict the normalized floating-point traffic flow of the rectangular grid G in Q consecutive time intervals in the future.

Optionally, performing simplified processing on the trajectory point sequence of the vehicle driving path according to the rectangular grid to obtain the gridded trajectory of the vehicle driving path, specifically including:

Each vehicle travel path is described by a sequence of track points, and each track point contains the longitude and latitude coordinates of the track point;

For a given vehicle driving path, the corresponding trajectory sequence is T, and the trajectory sequence T is described by N trajectory points: T={(x ₁ ,y ₁ ),(x ₂ ,y ₂ ),…,(x _N ,y _N )};

Among them, x ₁ and y ₁ represent the longitude and latitude of the first track point, respectively, and x _N and y _N represent the longitude and latitude of the Nth track point, respectively;

According to the rectangular grid G, the N trajectory points of the trajectory sequence T are divided into different segments in sequence, so that the trajectory points in each segment are located in the same grid area of the rectangular grid G:

第二个片段包含从第d₁+1个轨迹点

到第d₂个轨迹点

以此类推，第n个片段包含从第d_n-1+1个轨迹点

到第N个轨迹点(x_N,y_N)，| |表示片段间隔符；d₁、d₂、d_n-1、n均为自然数；Among them, let the trajectory T be divided into n segments, the first segment contains from the first trajectory point (x ₁ , y ₁ ) to the d _1th trajectory point

The second fragment contains ₁ +1 trajectory points from the dth

to the d _2nd trajectory point

And so on, the nth segment contains the trajectory points from d _n-1 +1

To the Nth track point (x _N , y _N ), | | represents the segment spacer; d ₁ , d ₂ , d _n-1 , and n are all natural numbers;

Calculate the average of the longitude and latitude of the trajectory points within the same segment to get the gridded trajectory T _G :

表示属于第i个片段内的轨迹点的经度的平均值，

表示属于第i个片段内的轨迹点的纬度的均值，x_j和y_j分别为经度坐标和纬度坐标，当i＝1时，d₀＝1；当i＝n时，d_n＝N；

in,

represents the mean of the longitudes of the trajectory points belonging to the ith segment,

Represents the mean value of the latitude of the trajectory points belonging to the ith segment, x _j and y _j are the longitude and latitude coordinates respectively, when i=1, d ₀ =1; when i=n, d _n =N;

Optionally, according to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the feature vector of each trajectory point in the gridded trajectory, specifically including:

Take out consecutive Q normalized floating-point traffic flow matrices Y _p+1 , Y _p+2 , . . . , Y _p+Q from the normalized floating-point traffic flow matrix; The sequence number of the time interval of a trajectory point (x ₁ , y ₁ ) in the time interval to which the training period belongs;

位于矩形网格G的第r_i行和第c_i列所在的格子区域，分别取出归一化浮点交通流量矩阵Y_p+1,Y_p+2,…,Y_p+Q中第r_i行和第c_i列的元素：

其中，

为归一化浮点交通流量矩阵Y_p+Q的第r_i行和第c_i列的元素；r_i和c_i分别为网格化轨迹T_G的第i个轨迹点

位于矩形网格的行序号和列序号；According to the i-th trajectory point of the gridded trajectory T _G

In the grid area where the ri- _th row and the c- _th column of the rectangular grid G are located, take out the _ri -th in the normalized floating-point traffic flow matrix Y _p+1 , Y _p+2 ,...,Y _p+Q respectively Elements of row and column c _i :

in,

are the elements of the ri-th row and ci- _th column of the normalized floating-point traffic flow matrix Y _p+Q ; _ri and _ci are the _ith trajectory point of the gridded trajectory T _G respectively

The row and column numbers in the rectangular grid;

Construct the eigenvector f _i of the ith trajectory point of the gridded trajectory T _G :

Among them, n is the number of trajectory points of the gridded trajectory T _G ; the superscript T represents the vector transposition.

Optionally, according to the feature vector of each trajectory point in the gridded trajectory, train the urban traffic travel time prediction network, specifically including:

Obtain the recording time o ₁ , o ₂ , . . . , o _N of each track point in the track sequence T of the vehicle travel path; wherein, o _N is the recording time of the Nth track point;

According to the rectangular grid G, each recording time is divided into different segments in sequence, so that the recording time in each segment is located in the same grid area of the rectangular grid G:

第二个片段包含从第d₁+1个记录时间

到第d₂个记录时间

以此类推，第n个片段包含从第d_n-1+1个记录时间

到第N个记录时间o_N，| |表示片段间隔符；d₁、d₂、d_n-1、n均为自然数| |表示分片间隔符，d₁、d₂、d_n-1、n均为自然数；Among them, it is assumed that the recording time will be divided into n segments, and the first segment contains the recording time from the first recording time o ₁ to the d ₁ th recording time

The second segment contains the record time from d ₁ + 1

to the d _2nd record time

And so on, the nth segment contains the record time from dn _-1 +1th

To the Nth recording time o _N , | | represents the segment separator; d ₁ , d ₂ , d _n-1 , and n are all natural numbers | | represents the fragment separator, d ₁ , d ₂ , d _n-1 , n is a natural number;

Calculate the travel time required to complete the current segment i:

为行驶完第i个片段的时间，对应于特征向量f_i的标记时间；i、j、d_i、d_i-1均为自然数，当i＝1时，d₀＝1；当i＝n时，d_n+1＝N；in,

is the time to complete the i-th segment, corresponding to the marked time of the feature vector f _i ; i, j, d _i , and d _i-1 are all natural numbers, when i=1, d ₀ =1; when i=n , d _n+1 =N;

According to the feature vector and the travel time of each segment, construct a second training sample:

为第二训练样本F的样本标记；f_n为车辆行驶路径的网格化轨迹的第n个轨迹点的特征向量；

为对应于特征向量f_n的标记时间；Among them, F represents the second training sample, the feature vector sequence f ₁ , f ₂ ..., f _n is the sample feature of the second training sample F, time series

is the sample label of the second training sample F; f _n is the feature vector of the nth trajectory point of the gridded trajectory of the vehicle travel path;

is the marked time corresponding to the feature vector f _n ;

According to the second training sample, an urban traffic travel time prediction network is trained, and the urban traffic travel time prediction network is used to predict the travel time of the entire travel path.

Optionally, based on the urban intra-city traffic flow prediction network and the urban intra-city travel time prediction network, and according to the to-be-tested travel path of the to-be-tested vehicle, determine the travel required by the to-be-tested vehicle to complete the to-be-tested path. time, including:

Taking the current time of the vehicle to be tested as the end time interval, calculate the normalized floating-point traffic flow matrix in the first J continuous time intervals, and stack each matrix into a three-dimensional tensor in chronological order:

A=[Y ₁ ,Y ₂ ,...,Y _J ];

Among them, A is a three-dimensional tensor composed of three-dimensional stereo data of L rows, M columns, and J layers. The normalized floating-point traffic flow matrices obtained in the first J consecutive time intervals with the current time interval as the end point are stacked in chronological order. , Y ₁ is the normalized floating-point traffic flow matrix obtained in the first J-1 time interval, Y ₂ is the normalized floating-point traffic flow matrix obtained in the first J-2 time interval, Y _J is the normalized floating-point traffic flow matrix obtained at the current time; L is the number of rows in the latitude direction of the rectangular grid G, and M is the number of columns in the longitude direction of the rectangular grid G;

According to the three-dimensional tensor A and the urban traffic flow prediction network, predict the normalized floating-point traffic flow in consecutive Q time intervals in the future: B ₁ , B ₂ , . . . , B _Q ;

Among them, B _Q is the normalized floating-point traffic flow in the Qth time interval in the future predicted by the urban traffic flow prediction network;

Determine the corresponding gridded trajectory R _G and the feature vector of each trajectory point according to the trajectory sequence R of the to-be-tested driving path of the vehicle to be tested;

Sample feature B of the trajectory sequence R: B=(b ₁ ,b ₂ ,...,b _n );

Among them, the sample feature B of the driving path R to be tested consists of a sequence of feature vectors b ₁ , b ₂ ,..., _bn ; b ₁ is the feature vector of the first track point of the gridded track R _G , and _bn is the grid The feature vector of the nth trajectory point of the gridded trajectory _RG ; n is the number of trajectory points contained in the gridded trajectory _RG ;

The eigenvector of the i-th trajectory point of the gridded trajectory R _G of the vehicle's driving path R:

和

分别表示网格化轨迹R_G的第i个轨迹点的经度和纬度，

为当前时刻城市市内交通流量预测网络所预测的第一个时间区间内的浮点归一化交通流量矩B₁的第r_i行和第c_i列的元素，

为当前时刻城市市内交通流量预测网络所预测的第二个时间区间内的浮点归一化交通流量矩阵B₂的第r_i行和第c_i列的元素，

为当前时刻城市市内交通流量预测网络所预测的第Q个时间区间内的浮点归一化交通流量矩阵B_Q的第r_i行和第c_i列的元素；r_i和c_i分别为网格化轨迹R_G的第i个轨迹点

位于矩形网格的行序号和列序号；Q为城市市内交通流量预测网络所预测的时间区间的个数，上标T表示向量转置；Among them, b _i represents the eigenvector of the i-th trajectory point of the gridded trajectory R _G ,

and

represent the longitude and latitude of the i-th track point of the gridded track R _G , respectively,

is the element of the r _i row and c _i column of the floating point normalized traffic flow moment B ₁ in the first time interval predicted by the urban traffic flow prediction network at the current moment,

is the element of row _ri and column _ci of the floating-point normalized traffic flow matrix B ₂ in the second time interval predicted by the urban traffic flow prediction network at the current moment,

is the element of the _{rith row and cith column} of the floating-point normalized traffic flow matrix B _Q in the Qth time interval predicted by the urban traffic flow prediction network at the current moment; _ri and _ci are respectively The ith trajectory point of the gridded trajectory R _G

The row number and column number located in the rectangular grid; Q is the number of time intervals predicted by the urban traffic flow prediction network, and the superscript T represents the vector transposition;

According to the eigenvectors of each track point of the gridded track R _G and the urban travel time prediction network, determine the time when the vehicle to be tested is driven on the travel path R to be tested.

In order to solve the above-mentioned technical problems, the present invention also provides the following solutions:

An urban traffic travel time prediction system, the prediction system includes:

The grid division unit is used to divide the city to be tested into rectangular grids;

a normalization processing unit for constructing a normalized floating-point traffic flow matrix according to historical traffic flow data based on the rectangular grid;

a first training unit, configured to train an urban traffic flow prediction network according to the normalized floating-point traffic flow matrix;

a simplification processing unit, configured to perform simplified processing on the trajectory point sequence of the vehicle driving path according to the rectangular grid to obtain the gridded trajectory of the vehicle driving path;

a feature determination unit, configured to determine the feature vector of each trajectory point in the gridded trajectory according to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path;

The second training unit is used for training the urban traffic travel time prediction network according to the feature vector of each trajectory point in the gridded trajectory;

The time prediction unit is used to determine, based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the travel path to be tested of the vehicle to be tested, what is required for the vehicle to be tested to complete the path to be tested travel time.

In order to solve the above-mentioned technical problems, the present invention also provides the following solutions:

An urban traffic travel time prediction system, comprising:

processor; and

memory arranged to store computer-executable instructions which, when executed, cause the processor to:

Divide the city to be tested into a rectangular grid;

Based on the rectangular grid, construct a normalized floating-point traffic flow matrix according to historical traffic flow data;

According to the normalized floating-point traffic flow matrix, train the urban traffic flow prediction network;

Simplify the trajectory point sequence of the vehicle's driving path according to the rectangular grid, and obtain the gridded trajectory of the vehicle's driving path;

According to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the eigenvectors of each trajectory point in the gridded trajectory;

According to the feature vector of each trajectory point in the gridded trajectory, train the urban traffic travel time prediction network;

Based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the travel path of the vehicle to be tested, the travel time required for the vehicle to be tested to complete the path to be tested is determined.

In order to solve the above-mentioned technical problems, the present invention also provides the following solutions:

A computer-readable storage medium storing one or more programs that, when executed by an electronic device including a plurality of application programs, cause the electronic device to perform the following operations :

Divide the city to be tested into a rectangular grid;

Based on the rectangular grid, construct a normalized floating-point traffic flow matrix according to historical traffic flow data;

According to the normalized floating-point traffic flow matrix, train the urban traffic flow prediction network;

Simplify the trajectory point sequence of the vehicle's driving path according to the rectangular grid, and obtain the gridded trajectory of the vehicle's driving path;

According to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the eigenvectors of each trajectory point in the gridded trajectory;

According to the feature vector of each trajectory point in the gridded trajectory, train the urban traffic travel time prediction network;

Based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the travel path of the vehicle to be tested, the travel time required for the vehicle to be tested to complete the path to be tested is determined.

According to the embodiments of the present invention, the present invention discloses the following technical effects:

The present invention divides the city to be tested into rectangular grids and refers to the historical traffic flow data to train the urban traffic flow prediction network and the urban traffic travel time prediction network. The time prediction network can determine the travel time of the vehicle to be tested in real time, which can improve the prediction accuracy in complex scenarios.

Description of drawings

Fig. 1 is the flow chart of the prediction method of the urban traffic travel time of the present invention;

Figure 2 is a structural diagram of a city traffic flow forecast network;

Figure 3 is a structural diagram of a city traffic travel time prediction network;

FIG. 4 is a schematic diagram of the module structure of the urban traffic travel time prediction system of the present invention.

Symbol Description:

Grid division unit-1, normalization processing unit-2, first training unit-3, simplification processing unit-4, feature determination unit-5, second training unit-6, temporal prediction unit-7.

Detailed ways

Preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principle of the present invention, and are not intended to limit the protection scope of the present invention.

The purpose of the present invention is to provide a method for predicting urban traffic travel time, by dividing the city to be measured into a rectangular grid, and referring to historical traffic flow data, to train the urban traffic flow prediction network and the urban traffic travel time prediction The network, through the urban traffic flow prediction network and the urban traffic travel time prediction network, can determine the travel time of the vehicle to be tested in real time, which can improve the prediction accuracy in complex scenarios.

In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

As shown in Figure 1, the method for predicting the travel time of urban traffic in the present invention includes:

Step 100: Divide the city to be tested into rectangular grids.

Step 200: Based on the rectangular grid, construct a normalized floating-point traffic flow matrix according to historical traffic flow data.

Step 300: Train an urban traffic flow prediction network according to the normalized floating-point traffic flow matrix.

Step 400: Simplify the trajectory point sequence of the vehicle travel path according to the rectangular grid to obtain a gridded trajectory of the vehicle travel path.

Step 500: According to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the feature vector of each trajectory point in the gridded trajectory.

Step 600: According to the feature vector of each trajectory point in the gridded trajectory, train a city traffic travel time prediction network.

Step 700: Based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the travel path of the vehicle to be tested, determine the travel time required for the vehicle to be tested to complete the path to be tested .

Wherein, in step 100, the city to be tested is divided into a rectangular grid G according to factors such as road grades, road network density, traffic driving status and other factors in each area of the city, and each row and column of the rectangular grid G is The size does not have to be the same. And the rectangular grid G is divided into L rows in the latitude direction and M columns in the longitude direction. Here, L and M are both natural numbers. In this way, the rectangular grid G divides the urban area of interest into L×M grid areas, and each grid area corresponds to a geographically different rectangular area.

In step 200, the normalized floating-point traffic flow matrix is constructed according to the historical traffic flow data based on the rectangular grid, which specifically includes:

Step 210: selecting the historical n _d days as the training period, and dividing each day of the training period into K _d time intervals with fixed time intervals; the total number of the continuous time intervals that the training period is divided into is K: K=n _d ×K _d .

It should be pointed out that this involves a problem of geographical area division and time interval selection. If the grid area is too small, there will be some time intervals where there is no vehicle flow in some grid areas; if the grid area is too large, it is difficult to accurately describe the traffic flow characteristics. In the embodiment of the present invention, the number of rows L in the latitude direction of the rectangular grid G is set to 120, the number of columns M of the rectangular grid G in the longitude direction is 120, and the number of time intervals K _d divided every day is 360. That is, the time interval is 4 minutes. The time interval of 4 minutes is chosen mainly because for the vast majority of cities, the traffic state usually does not change frequently during this time interval.

Step 220: Count the trajectory data of the vehicles entering the networked system in the rectangular grid G during the training period, and each trajectory data records the longitude position, latitude position and time information of the vehicle.

When implementing the present invention, the networking system may be a combination of one or more systems, such as a taxi operation management system, a public transport vehicle bus operation management system, a traffic travel service system, and a car networking.

表示在第t个时间区间内出现在矩形网格G的第i行第j列格子区域内的车辆个数。这里

为自然数，随后，将序号i遍历1至L之间的整数，序号j遍历1至M之间的整数，得到一个大小为L行M列的流量矩阵。记该流量矩阵为X_t。流量矩阵X_t的每一个元素记录第t个时间区间内在对应格子区域内的车辆流量。t的取值范围为1至K之间的整数。

Indicates the number of vehicles that appear in the grid area of the i-th row and j-th column of the rectangular grid G in the t-th time interval. here

is a natural number, then, traverse the sequence number i through the integers between 1 and L, and the sequence number j through the integers between 1 and M, to obtain a flow matrix with a size of L rows and M columns. Let this flow matrix be X _t . Each element of the flow matrix X _t records the vehicle flow in the corresponding grid area in the t-th time interval. The value range of t is an integer between 1 and K.

Step 230: Traverse all the time intervals in the training period, count the traffic flow of vehicles entering the networked system in the same time interval, and obtain K traffic flow matrices, which are respectively denoted as X ₁ , X ₂ ,...,X _k , ...,X _K ; X _k represents the traffic flow matrix in the kth time interval, k=1,2,....,K.

Step 240: Perform normalized floating-point processing on each traffic flow matrix according to the following formula to obtain a corresponding normalized floating-point traffic flow matrix:

为第t个时间区间内的归一化浮点交通流量矩阵Y_t的第i行第j列元素，x₀表示K个交通流量矩阵中所有元素的最大值(出现在所有时间区间和所有格子区间内的车辆数目的最大值)。in,

is the i-th row and j-th column element of the normalized floating-point traffic flow matrix Y _t in the t-th time interval, x ₀ represents the maximum value of all elements in the K traffic flow matrices (appears in all time intervals and all grids maximum number of vehicles in the interval).

By introducing the floating point normalization operation, the relative value of the traffic flow of each rectangular grid can be provided, thereby reducing the deviation caused by incomplete traffic flow statistics.

In step 300, a training sample set required for the urban traffic flow prediction network is constructed. The task of the urban traffic flow prediction network is to use the time series data formed by the normalized floating-point traffic flow matrix to predict the traffic flow in multiple continuous time intervals in the future. In order to train the intra-city traffic flow prediction network, training samples need to be constructed from the floating-point normalized traffic flow matrix obtained during the training period. Each training sample consists of sample features and output labels.

Specifically, according to the normalized floating-point traffic flow matrix, the training of the urban traffic flow prediction network includes:

Step 310: According to the normalized floating-point traffic flow matrix, construct a corresponding first training sample:

P _t =(A _t ; B _t ),

Among them, t represents the sequence number of the time interval, P _t represents the first training sample; A _t represents the sample feature of the first training sample P _t , which is a three-dimensional tensor with L rows, M columns, and J layers, starting from the t-th time interval as the end point. The normalized traffic flow _matrices obtained in the _{first J consecutive} time _intervals of the ]; where Y _t-J+1 is the normalized floating-point traffic flow matrix obtained in the t+1th time interval, and Y _t-J+2 is the normalized floating-point traffic flow matrix obtained in the t+2th time interval Floating-point traffic flow matrix, Y _t-1 is the normalized floating-point traffic flow matrix obtained in the t-1th time interval, Y _t is the normalized floating-point traffic flow matrix obtained in the t-th time interval, The natural number L is the number of rows in the latitude direction of the rectangular grid G, the natural number M is the number of columns in the longitude direction of the rectangular grid G, J is the number of continuous time intervals for sample features, and Q is the number of continuous time intervals for constructing sample markers (that is, the number of normalized floating-point traffic flow matrices in the future continuous time interval that the urban traffic flow forecasting network needs to predict). Here, J and Q are both natural numbers;

B _t represents the sample label of the first training sample P _t , which is a three-dimensional tensor with L rows and M columns and Q layers. The normalized traffic obtained in Q consecutive time intervals starting from the t+1th time interval The flow matrix is stacked in chronological order: B _t =[Y _t+1 ,Y _t+2 ,…,Y _t+Q ];

Among them, Y _t+1 is the normalized floating-point traffic flow matrix obtained in the t+1 time interval, and Y _t+2 is the normalized floating-point traffic flow obtained in the t+2 time interval Matrix, Y _t+Q is the normalized floating-point traffic flow matrix obtained in the t+Qth time interval, the natural number L is the row number of the rectangular grid G in the latitude direction, and the natural number M is the rectangular grid G in the longitude direction The number of columns of , the natural number Q is the number of continuous time intervals for constructing sample markers, and t is a natural number. Both the sample feature _At and the sample label _Bt are known.

Considering the need to predict the traffic data in Q consecutive time intervals in the future, for a normalized floating-point traffic flow matrix with a total length of K, t can take an integer between 1 and K-J-Q+1. Therefore, a total of K-J-Q training samples can be obtained in the above manner. In this embodiment, the natural numbers J and Q are both set to 10.

Step 320: Train an urban traffic flow prediction network according to the first training sample; the urban urban traffic flow prediction network is used to predict the normalized floating-point traffic of the rectangular grid G in Q consecutive time intervals in the future flow.

The backbone structure of the urban traffic flow prediction network in this embodiment is shown in Table 1. In Table 1, conv stands for convolutional layer. The urban traffic flow prediction network used in this embodiment has a total of 7 convolution layers. Convolutional layer conv1, convolutional layer conv2 and convolutional layer conv3 form an encoder for learning high-level semantic features; convolutional layer conv4, convolutional layer conv5, convolutional layer conv6 and convolutional layer conv7 form a decoder for The traffic flow prediction results in each grid area are obtained by step-by-step decoding from high-level semantic features. The size of the convolution kernel of the urban traffic flow prediction network in this embodiment is all 3×3.

In Table 1, the number of input channels of the convolutional layer conv1 is 10. This is because the normalized floating-point traffic flow matrix obtained in the urban traffic flow prediction network used in this embodiment is obtained in 10 consecutive time regions as Input data. The number of input channels of the convolutional layer conv7 is 10, because the urban traffic flow prediction network used in this embodiment uses the normalized floating-point traffic flow data predicted in 10 continuous time intervals in the future as the output.

In Table 1, the output of the convolutional layer conv1, convolutional layer conv2, convolutional layer conv3, convolutional layer conv4, convolutional layer conv5, and convolutional layer conv6 of the urban traffic flow prediction network used in this embodiment The number of channels is 20, 40, 80, 40, 20, 10, respectively. Therefore, the parameters in Table 1 determine the structure of the urban traffic flow prediction network used in this embodiment.

Each convolution operation is followed by an excitation operation, which is performed by the commonly used Rectified Linear Unit (ReLU). The linear rectification function is to take the maximum value of the result of the convolution and zero, that is, if the result of the convolution is greater than zero, the result is retained, otherwise it is set to zero.

It is worth noting that, in the present invention, the intra-city traffic flow prediction network does not perform the commonly used pooling operation in the encoder, that is, the down-sampling operation of the rectangular grid area. This is because, compared with the global (the whole city), the urban traffic conditions reflect more local regional correlations, so it is not necessary to expand the region of feature learning through downsampling.

As shown in Figure 2, the output of the convolutional layer conv1 (after the excitation operation) is the input of the convolutional layer conv2, the output of the convolutional layer conv2 (after the excitation operation) is the input of the convolutional layer conv3, and the convolutional layer conv3 The output of (after excitation operation) is the input of the convolutional layer conv4.

As shown in Figure 2, the input of the convolutional layer conv5 is formed by concatenating the output of the convolutional layer conv2 and the output of the convolutional layer conv4, that is, the "conc" operation shown in Figure 2. In this way, the number of input channels of the convolutional layer conv5 is 80, as shown in Table 1. The input of the convolutional layer conv6 is the concatenation of the output of the convolutional layer conv1 and the output of the convolutional layer conv5, that is, the "concatenation" operation shown in Figure 2. In this way, the number of input channels of the convolutional layer conv6 is 40, as shown in Table 1. The input of the convolutional layer conv7 is the concatenation of the input of the convolutional layer conv1 and the output of the convolutional layer conv6, that is, the "concatenation" operation shown in Figure 2. In this way, the number of input channels of the convolutional layer conv7 is 20, as shown in Table 1. In the present invention, the "concatenation" operation is introduced to increase the precision of the normalized floating point traffic flow within each grid area.

Table 1: The size of the convolution kernel of the urban traffic flow prediction network and the number of input and output channels in each layer

name convolution kernel size Number of input channels Number of output channels conv1 3×3 10 20 conv2 3×3 20 40 conv3 3×3 40 80 conv4 3×3 80 40 conv5 3×3 80 20 conv6 3×3 40 20 conv7 3×3 30 10

Finally, the urban traffic flow prediction network used in this embodiment is trained. In the present invention, the mean squared error (Mean Squared Error, MSE) is used as the loss function of the network. The optimization algorithm adopts the classic error back-propagation algorithm in the field. In the present invention, the size of the learning rate is 0.001, the size of the batch data set is set to 16, and each batch data set is trained for 20 rounds in total.

After the network is trained, when the network is applied, given a 3D tensor stacked in chronological order from normalized floating-point traffic flow matrices obtained in consecutive J time regions, the network will output a 3D tensor It consists of stacking Q matrices, each of which is a normalized floating-point traffic flow matrix in a future time interval predicted by the network in chronological order.

The present invention divides the urban area of interest into a rectangular grid G containing L rows and M columns. The spatial resolution of the rectangular grid G is smaller than the spatial resolution of the vehicle travel path that can be recorded by the existing positioning system (such as the north navigation system). In order to characterize the travel path, it is necessary to keep the trajectory point sequence of the travel path consistent with the spatial resolution of the rectangular grid G. Therefore, it is necessary to perform grid simplification on the original trajectory sequence of the vehicle travel path according to the rectangular grid G.

Specifically, in step 400, a simplified process is performed on the trajectory point sequence of the vehicle's driving path according to the rectangular grid to obtain a gridded trajectory of the vehicle's driving path, including:

Step 410: Each vehicle travel path is described by a sequence of track points, and each track point includes the longitude and latitude coordinates where the track point is located;

For a given vehicle driving path, the corresponding trajectory sequence is T, and the trajectory sequence T is described by N trajectory points: T={(x ₁ ,y ₁ ),(x ₂ ,y ₂ ),…,(x _N ,y _N )};

Among them, x ₁ and y ₁ represent the longitude and latitude of the first track point, respectively, and x _N and y _N represent the longitude and latitude of the Nth track point, respectively.

Step 420: According to the rectangular grid G, divide the N trajectory points of the trajectory sequence T into different segments in sequence, so that the trajectory points in each segment are located in the same grid area of the rectangular grid G:

第二个片段包含从第d₁+1个轨迹点

到第d₂个轨迹点

以此类推，第n个片段包含从第d_n-1+1个轨迹点

到第N个轨迹点(x_N,y_N)，| |表示片段间隔符；d₁、d₂、d_n-1、n均为自然数。Among them, let the trajectory T be divided into n segments, the first segment contains from the first trajectory point (x ₁ , y ₁ ) to the d _1th trajectory point

The second fragment contains ₁ +1 trajectory points from the dth

to the d _2nd trajectory point

And so on, the nth segment contains the trajectory points from d _n-1 +1

To the Nth track point (x _N , y _N ), | | represents the segment spacer; d ₁ , d ₂ , d _n-1 , and n are all natural numbers.

Step 430: Calculate the average of the longitude and latitude of the track points in the same segment to obtain the gridded track T _G :

表示属于第i个片段内的轨迹点的经度的平均值，

表示属于第i个片段内的轨迹点的纬度的均值，x_j和y_j分别为经度坐标和纬度坐标，当i＝1时，d₀＝1；当i＝n时，d_n＝N；

in,

represents the mean of the longitudes of the trajectory points belonging to the ith segment,

Represents the mean value of the latitude of the trajectory points belonging to the ith segment, x _j and y _j are the longitude and latitude coordinates respectively, when i=1, d ₀ =1; when i=n, d _n =N;

The symbol "x" represents the longitude coordinate of the corresponding trajectory point, the symbol "y" represents the latitude coordinate of the corresponding trajectory point, and d ₁ , d ₂ , d _n-1 , and n are all natural numbers.

For a given vehicle travel path, it is not enough to use only the position coordinates of each track point in the path to predict the time to complete the track. To achieve this purpose, the present invention adds the normalized floating point traffic flow in the corresponding grid area to each track point. Normalized floating-point traffic flow is dynamic information that is closely related to vehicle travel time and can therefore be used to estimate travel time.

位于矩形网格G的第r_i行和第c_i列所在的格子区域。在这里，自然数i的取值范围为1到n之间的整数，自然数r_i的取值范围为1至L之间的整数，自然数c_i的取值范围为1至M之间的整数，自然数L为矩形网格G在纬度方向的行数，自然数M为矩形网格G在经度方向的列数。Assume that the ith trajectory point of the gridded trajectory T _G obtained after the vehicle travel path trajectory T is simplified by the gridding described in step S3

It is located in the grid area where the ri- _th row and the ci- _th column of the rectangular grid G are located. Here, the value range of the natural number _i is an integer between 1 and n, the value range of the natural number ri is an integer between 1 and L, and the value range of the natural number c _i is an integer between 1 and M, The natural number L is the number of rows of the rectangular grid G in the latitude direction, and the natural number M is the number of columns of the rectangular grid G in the longitude direction.

Specifically, in step 500, according to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the feature vector of each trajectory point in the gridded trajectory, including:

Step 510: Take out consecutive Q normalized floating-point traffic flow matrices Y _p+1 , Y _p+2 , . . . , Y _p+Q from the normalized floating-point traffic flow matrix; The sequence number of the time interval of the first trajectory point (x ₁ , y ₁ ) of the path T in the time interval to which the training period belongs.

During the training period, since the vehicle travel path trajectory T belongs to historical data, the time of its first trajectory point (x ₁ , y ₁ ) is known. The index value of the time interval to which the time of the first trajectory point (x ₁ , y ₁ ) belongs to the training period in step S1 is recorded as p. Here p is a natural number.

位于矩形网格G的第r_i行和第c_i列所在的格子区域，分别取出归一化浮点交通流量矩阵Y_p+1,Y_p+2,…,Y_p+Q中第r_i行和第c_i列的元素：

其中，

为归一化浮点交通流量矩阵Y_p+Q的第r_i行和第c_i列的元素；r_i和c_i分别为网格化轨迹T_G的第i个轨迹点

位于矩形网格的行序号和列序号。Step 520: According to the i-th trajectory point of the gridded trajectory T _G

In the grid area where the ri- _th row and the c- _th column of the rectangular grid G are located, take out the _ri -th in the normalized floating-point traffic flow matrix Y _p+1 , Y _p+2 ,...,Y _p+Q respectively Elements of row and column c _i :

in,

are the elements of the ri-th row and ci- _th column of the normalized floating-point traffic flow matrix Y _p+Q ; _ri and _ci are the _ith trajectory point of the gridded trajectory T _G respectively

The row and column numbers in the rectangular grid.

Step 530: Construct the feature vector f _i of the i-th trajectory point of the gridded trajectory _TG :

和

分别表示网格化轨迹T_G的第i个轨迹点的经度和纬度，

为浮点归一化交通流量矩阵Y_p+1的第r_i行和第c_i列的元素，

为浮点归一化交通流量矩阵Y_p+2的第r_i行和第c_i列的元素，

为浮点归一化交通流量矩阵Y_p+Q的第r_i行和第c_i列的元素；r_i和c_i分别为网格化轨迹T_G的第i个轨迹点

位于矩形网格的行索引值和列索引值；p为车辆行驶路径T第一个轨迹点的时间在训练时段所属时间区间的索引值；自然数Q为城市市内交通流量预测网络所预测的时间区间的个数；自然数n为网格化轨迹T_G的轨迹点的个数；自然数i的取值范围为1到n之间的整数；上标T表示向量转置。Among them, f _i represents the feature vector of the i-th trajectory point of the gridded trajectory T _G ,

and

respectively represent the longitude and latitude of the i-th track point of the gridded track T _G ,

is the floating point normalized traffic flow matrix Y _p+1 elements in row r _i and column c _i ,

is the floating point normalized traffic flow matrix Y _p+2 elements in row r _i and column c _i ,

is the element of the ri-th row and ci- _th column of the floating-point normalized traffic flow matrix Y _p+Q ; _ri and _ci are the _ith trajectory point of the gridded trajectory T _G respectively

The row index value and column index value located in the rectangular grid; p is the index value of the time of the first trajectory point of the vehicle travel path T in the time interval of the training period; the natural number Q is the time predicted by the urban traffic flow prediction network The number of intervals; the natural number n is the number of trajectory points of the gridded trajectory T _G ; the value range of the natural number i is an integer between 1 and n; the superscript T represents the vector transposition.

Each trajectory point of the gridded trajectory T _G has both the position coordinate information and the dynamic information related to the traffic state, which is beneficial to improve the accuracy of the travel time estimation.

In step 600, according to the feature vector of each trajectory point in the gridded trajectory, train the urban traffic travel time prediction network, which specifically includes:

Step 610: Acquire the recording time o ₁ , o ₂ , . . . , o _N of each track point in the track sequence T of the vehicle traveling path; wherein, o _N is the recording time of the Nth track point.

Specifically, for a trajectory T of a vehicle traveling path located in a rectangular grid G, the gridded trajectory T _G is obtained by performing grid simplification on the trajectory T according to the method of step 400 . Further, according to the method of step 500, the feature vector of each trajectory point of the gridded trajectory T _G is obtained, that is, f ₁ , f ₂ . . . , f _n . Among them, f ₁ is the eigenvector of the first trajectory point of the gridded trajectory _TG , f2 is the eigenvector of the _second trajectory point of the gridded trajectory _TG , and _fn is the eigenvector of the gridded trajectory _TG The eigenvector of the nth trajectory point. Here, the natural number n is the number of trajectory points included in the gridded trajectory _TG .

Correspondingly, for the trajectory T of the vehicle travel path, the number of trajectory points contained in it is denoted as N. The recording time of each track point of the track path is retrieved from the database, and is arranged in sequence as o ₁ , o ₂ ,...,o _N . Among them, o ₁ is the recording time of the first track point of the track T, o ₂ is the recording time of the first track point, and o _N is the recording time of the Nth track point. Therefore, the total time to travel the trajectory T is o _N -o ₁ .

Step 620: According to the rectangular grid G, divide each recording time into different segments in sequence, so that the recording time in each segment is located in the same grid area of the rectangular grid G:

第二个片段包含从第d₁+1个记录时间

到第d₂个记录时间

以此类推，第n个片段包含从第d_n-1+1个记录时间

到第N个记录时间o_N，| |表示片段间隔符；d₁、d₂、d_n-1、n均为自然数| |表示分片间隔符，d₁、d₂、d_n-1、n均为自然数。Among them, it is assumed that the recording time will be divided into n segments, and the first segment contains the recording time from the first recording time o ₁ to the d ₁ th recording time

The second segment contains the record time from d ₁ + 1

to the d _2nd record time

And so on, the nth segment contains the record time from dn _-1 +1th

To the Nth recording time o _N , | | represents the segment separator; d ₁ , d ₂ , d _n-1 , and n are all natural numbers | | represents the fragment separator, d ₁ , d ₂ , d _n-1 , n are all natural numbers.

Since the driving paths of the vehicles are different, the number of their trajectory points is also different. That is to say, for different tracks T, the number N of track points is different. Correspondingly, the number n of track points of the gridded track T _G is also different. In order to adapt to the situation of the different lengths of the trajectory, the present invention will output a travel time in each grid area along the gridded trajectory _TG .

Step 630: Calculate the travel time required to complete the current segment i:

为行驶完第i个片段的时间，对应于特征向量f_i的标记时间；i、j、d_i、d_i-1均为自然数，当i＝1时，d₀＝1；当i＝n时，d_n+1＝N。in,

is the time to complete the i-th segment, corresponding to the marked time of the feature vector f _i ; i, j, d _i , and d _i-1 are all natural numbers, when i=1, d ₀ =1; when i=n , d _n+1 =N.

Step 640: Construct a second training sample according to the feature vector and the travel time of each segment:

为第二训练样本F的样本标记；f₁为车辆行驶路径的网格化轨迹的第一个轨迹点的特征向量，f₂为车辆行驶路径的网格化轨迹的第二个轨迹点的特征向量，f_n为车辆行驶路径的网格化轨迹的第n个轨迹点的特征向量；

为对应于特征向量f₁的标记时间，

为对应于特征向量f₂的标记时间，

为对应于特征向量f_n的标记时间；自然数n为车辆行驶路径的网格化轨迹的轨迹点的个数。Among them, F represents the second training sample, the feature vector sequence f ₁ , f ₂ ..., f _n is the sample feature of the second training sample F, time series

is the sample label of the second training sample F; f1 is the feature vector of the _first trajectory point of the gridded trajectory of the vehicle driving path, and f2 is the feature of the _second trajectory point of the gridded trajectory of the vehicle driving path vector, f _n is the feature vector of the nth trajectory point of the gridded trajectory of the vehicle travel path;

is the labeling time corresponding to the feature vector f ₁ ,

is the labeling time corresponding to the feature vector _f2 ,

is the marked time corresponding to the feature vector f _n ; the natural number n is the number of trajectory points of the gridded trajectory of the vehicle travel path.

Step 650: Train an urban traffic travel time prediction network according to the second training sample, where the urban urban traffic travel time prediction network is used to predict the travel time of the entire travel path.

As shown in FIG. 3 , the present invention uses a standard Long-Short Term Memory (LSTM) unit as the information processing unit at each moment to design the main structure of the network.

接着，将特征向量f₂和隐状态向量v₁同时送入至第二个轨迹点对应的LSTM单元，该单元输出一个隐状态向量v₂，进一步将隐状态向量v₂输入至FCN单元，输出行驶完前两个片段的总的行驶时间，记该为预测时间

以此类推，最后将特征向量f_n送入至第n个轨迹点对应的LSTM单元，通过FCN单元输出行驶完前n个片段的总的行驶时间，即整个轨迹的行驶时间，记该为预测时间

For a training sample F, the feature vector f ₁ is first sent to the LSTM unit corresponding to the first trajectory point, which outputs a hidden state vector v ₁ , and further inputs the hidden state vector v ₁ to a fully connected neural network ( Fully-Connected Network, FCN). Denote the fully connected neural network as the FCN unit. _The FCN unit takes the hidden state vector v1 as the input layer, there is no hidden layer, the output layer contains a node, and the excitation function of the output layer is a linear rectification function, that is, the ReLU used in the step. At this time, the FCN unit outputs the predicted time when the first trajectory segment is completed, which is recorded as the predicted time.

Next, the feature vector f ₂ and the hidden state vector v ₁ are simultaneously sent to the LSTM unit corresponding to the second trajectory point, the unit outputs a hidden state vector v ₂ , and the hidden state vector v ₂ is further input to the FCN unit, output The total driving time after driving the first two segments is recorded as the predicted time

By analogy, the feature vector f _{n is} finally sent to the LSTM unit corresponding to the nth trajectory point, and the total travel time of the first n segments is output through the FCN unit, that is, the travel time of the entire trajectory, which is recorded as prediction. time

It should be pointed out that the above processing process can be applied to training samples containing any number of trajectory points. The number of trajectory points that can be processed by the urban traffic travel time prediction network in the embodiment of the present invention is designed according to the maximum number of trajectory points contained in all gridded trajectories _TG in the training sample set.

其中[]代表取整运算。在这里，Q为步骤300中城市市内交通流量预测网络所输出的连续时间区间交通流量矩阵的个数。In the above processing process, according to the calculation process of the standard LSTM unit, only the dimension of the hidden state vector of the LSTM unit needs to be determined to determine the structure of the entire network. In the present invention, the dimension of the hidden state vector is set as half of the dimension of the feature vector f ₁ , f ₂ . . . , f _n , that is,

Where [] represents the rounding operation. Here, Q is the number of continuous time interval traffic flow matrices output by the urban traffic flow prediction network in step 300 .

In the whole urban traffic travel time prediction network, the structure of all LSTM units is the same, and the parameters are also the same. In addition, the structure of all FCN units is the same, and the parameters are also the same. The advantage of this design is that when applying the network, the network can be extended by copying the LSTM unit and the FCN unit, so that it can be applied to driving trajectories of any length.

Next, define the loss function. For each trajectory point of the gridded trajectory _TG , the fully connected neural network P outputs a prediction time, so a loss function needs to be defined. The loss caused by sample F is calculated by the following loss function:

为FCN单元输出的网格化轨迹T_G的前i个片段的预测时间，

为样本F的已知的标记时间。在这里，对样本F，自然数i的取值范围为1到n之间的整数。Among them, loss(F) represents the loss caused by sample F,

is the predicted time of the first i segments of the gridded trajectory T _G output by the FCN unit,

is the known marking time of sample F. Here, for the sample F, the value range of the natural number i is an integer between 1 and n.

Finally, train a travel time prediction network for intra-city traffic. In the present invention, the standard error back-propagation algorithm is used to train the intra-city travel time prediction network. During this process, in the embodiment of the present invention, the size of the learning rate is set to 0.001, the size of the batch data set is set to 64, and each batch data set is trained for a total of 20 rounds.

In step 700, based on the urban intra-city traffic flow prediction network and the urban intra-city travel time prediction network, and according to the to-be-tested travel path of the to-be-tested vehicle, determine the amount of time required for the tested vehicle to travel the to-be-tested path. Travel time, including:

Step 710: Take the current time when the vehicle to be tested is traveling as the end time interval, calculate the normalized floating-point traffic flow matrix in the first J consecutive time intervals, and stack each matrix into a three-dimensional tensor in chronological order:

A=[Y ₁ ,Y ₂ ,...,Y _J ];

Among them, A is a three-dimensional tensor composed of three-dimensional stereo data of L rows, M columns, and J layers. The normalized floating-point traffic flow matrices obtained in the first J consecutive time intervals with the current time interval as the end point are stacked in chronological order. , Y ₁ is the normalized floating-point traffic flow matrix obtained in the first J-1 time interval, Y ₂ is the normalized floating-point traffic flow matrix obtained in the first J-2 time interval, Y _J is the normalized floating-point traffic flow matrix obtained at the current time; L is the number of rows in the latitude direction of the rectangular grid G, and M is the number of columns in the longitude direction of the rectangular grid G.

Step 720: Predict the normalized floating-point traffic flow in consecutive Q time intervals in the future according to the three-dimensional tensor A and the urban traffic flow prediction network: B ₁ , B ₂ , . . . , B _Q ;

Among them, B ₁ is the normalized floating-point traffic flow in the first time interval in the future predicted by the urban traffic flow prediction network; B ₂ is the normalized traffic flow in the second time interval in the future predicted by the network Normalized floating-point traffic flow; B _Q is the normalized floating-point traffic flow predicted by the network in the Qth time interval in the future.

Step 730: Determine the corresponding gridded trajectory R _G and the feature vector of each trajectory point according to the trajectory sequence R of the driving path to be tested of the vehicle to be tested;

Sample feature B of the trajectory sequence R: B=(b ₁ ,b ₂ ,...,b _n );

Among them, the sample feature B of the driving path R to be tested consists of a sequence of feature vectors b ₁ , b ₂ ,..., _bn ; b ₁ is the feature vector of the first track point of the gridded track R _G , and b ₂ is the grid The eigenvector of the second trajectory point of the gridded trajectory _RG , b _n is the _eigenvector of the nth trajectory point of the gridded trajectory RG; the natural number n is the number of trajectory points contained in the gridded trajectory _RG number;

The eigenvector of the i-th trajectory point of the gridded trajectory R _G of the vehicle's driving path R:

和

分别表示网格化轨迹R_G的第i个轨迹点的经度和纬度，设网格化轨迹R_G的第i个轨迹点

位于矩形网格G的第r_i行和第c_i列所在的格子区域。在这里，自然数i的取值范围为1到n之间的整数；自然数r_i的取值范围为1至L之间的整数，自然数c_i的取值范围为1至M之间的整数；

为当前时刻城市市内交通流量预测网络所预测的第一个时间区间内的浮点归一化交通流量矩B₁的第r_i行和第c_i列的元素，

为当前时刻城市市内交通流量预测网络所预测的第二个时间区间内的浮点归一化交通流量矩阵B₂的第r_i行和第c_i列的元素，

为当前时刻城市市内交通流量预测网络所预测的第Q个时间区间内的浮点归一化交通流量矩阵B_Q的第r_i行和第c_i列的元素；r_i和c_i分别为网格化轨迹R_G的第i个轨迹点

位于矩形网格的行序号和列序号；自然数Q为城市市内交通流量预测网络所预测的时间区间的个数，上标T表示向量转置；Among them, b _i represents the eigenvector of the i-th trajectory point of the gridded trajectory R _G ,

and

respectively represent the longitude and latitude of the ith track point of the gridded track R _G , and set the ith track point of the grid track _RG

It is located in the grid area where the ri- _th row and the ci- _th column of the rectangular grid G are located. Here, the value range of the natural number _i is an integer between 1 and n; the value range of the natural number ri is an integer between 1 and L, and the value range of the natural number c _i is an integer between 1 and M;

is the element of the r _i row and c _i column of the floating point normalized traffic flow moment B ₁ in the first time interval predicted by the urban traffic flow prediction network at the current moment,

is the element of row _ri and column _ci of the floating-point normalized traffic flow matrix B ₂ in the second time interval predicted by the urban traffic flow prediction network at the current moment,

is the element of the _{rith row and cith column} of the floating-point normalized traffic flow matrix B _Q in the Qth time interval predicted by the urban traffic flow prediction network at the current moment; _ri and _ci are respectively The ith trajectory point of the gridded trajectory R _G

The row number and column number located in the rectangular grid; the natural number Q is the number of time intervals predicted by the urban traffic flow prediction network, and the superscript T represents the vector transposition;

Step 740 : According to the feature vector of each track point of the gridded track R _G and the urban travel time prediction network, determine the time when the vehicle to be tested is driven on the travel path R to be tested.

Specifically, the _{feature vectors} b ₁ , b ₂ , . The last fully connected neural network FCN unit outputs the time when the vehicle to be predicted travels the route R, which is the time predicted by the urban traffic flow prediction network.

The prediction method of the urban traffic travel time of the present invention can predict the urban traffic travel time of a given route, which is mainly reflected in the following aspects:

1) The present invention needs to train two networks, one network predicts real-time road condition information, and the other network uses the prediction result of the previous network to predict the urban traffic travel time of a given route. This framework is designed to take into account the real-time traffic conditions that affect the travel time of intra-city traffic.

2) The method of deep learning is used. Its excellent nonlinear fitting ability has been proven in many machine learning fields. It can achieve better results than traditional methods in urban traffic travel time prediction;

3) Although the real-time traffic flow information is used as the representative of the additional factors affecting the traffic travel time in the city, the framework of the present invention can be easily applied to other factors, and only needs to partially modify the network structure. To consider the impact of dynamically changing weather information on urban traffic travel time, it is only necessary to replace the urban traffic flow forecast network with a real-time weather forecast network. This increases the flexibility of the present invention.

The following experiments are carried out on the taxi trajectory data of Yimou City. In order to verify the effectiveness of the present invention, two trajectories are shown in the figure. Among them, under the trajectory sequence of the travel path R ₁ :

([104.048087,30.67257],[104.048688,30.667996],[104.046058,30.663864],[104.045486,30.663977],[104.041591,30.665894],[104.041185,30.665958],[104.045532,30.663658],[104.045747,30.660006],[ 104.046027,30.656351],[104.049302,30.653285],[104.052818,30.652634],[104.05679,30.649528],[104.060861,30.647632],[104.065253,30.647387],[104.068546,30.646833],[104.070377,30.646512],[104.074429, 30.646343],[104.07896,30.645559],[104.082964,30.643429],[104.086483,30.642236],[104.088603,30.639872],[104.087352,30.63816],[104.08751,30.638336],[104.088682,30.639902],[104.088278,30.640154] ,[104.087855,30.639184],[104.092022,30.641449],[104.094963,30.638586],[104.094659,30.637947],[104.09227,30.637359],[104.089681,30.640921],[104.086951,30.638889],[104.08346,30.637202],[ 104.07876,30.634991],[104.074695,30.633712],[104.069852,30.633636],[104.064814,30.633495],[104.060504,30.633411],[104.055255,30.63328],[104.050522,30.63545],[104.04863,30.63635],[104.046449, 30.637371],[104.042775,30.636046],[104.040166,30.631991] ,[104.037331,30.630241],[104.033926,30.633511],[104.030826,30.636792],[104.028128,30.639455],[104.025346,30.642233],[104.02234,30.641511],[104.018598,30.638672],[104.013982,30.63651],[ 104.01379,30.636413],[104.009396,30.634587],[104.006292,30.633066],[104.002995,30.631547],[103.999547,30.629926],[103.994709,30.627643],[103.990154,30.625509],[103.985626,30.623449],[103.981084, 30.621406], [103.97575757,30.619052], [103.972676,30.617613], [103.970359, 30.616433], [103.979211,33.3.33.3.33.333333]

The trajectory sequence of the travel path R ₂ is as follows:

([104.107954,30.694667],[104.107533,30.692622],[104.106701,30.6916],[104.105737,30.69041],[104.104835,30.689299],[104.103889,30.688151],[104.102697,30.686618],[104.102828,30.68506],[ 104.103921,30.683158],[104.104684,30.681414],[104.10577,30.679865],[104.107494,30.677033],[104.108455,30.675276],[104.109234,30.673234],[104.110213,30.671347],[104.110638,30.670777],[104.111095, 30.668693],[104.111861,30.666929],[104.112352,30.66585],[104.113085,30.663795],[104.113307,30.663361],[104.114126,30.661514],[104.114649,30.659982],[104.1149,30.657735],[104.114941,30.656178] ,[104.114834,30.654379],[104.114826,30.652352],[104.115693,30.650939],[104.117447,30.650225],[104.116893,30.648046],[104.118718,30.648134],[104.118882,30.64808],[104.118639,30.647728],[ 104.117693,30.647957],[104.116512,30.647634],[104.116169,30.646546],[104.115794,30.645291],[104.11504,30.643075],[104.11424,30.641665],[104.111912,30.642674],[104.110404,30.64393],[104.108867, 30.641481],[104.107777,30.63992],[104.106731,30.63847],[ 104.105625,30.636528],[104.104823,30.635327],[104.104013,30.634062],[104.102934,30.632328],[104.101653,30.630342],[104.100493,30.628484],[104.100305,30.62732],[104.099455,30.626544],[104.098251, 30.624969],[104.097108,30.623899],[104.095813,30.622056],[104.094057,30.621055],[104.091939,30.620839],[104.089401,30.620788],[104.087057,30.62065],[104.085334,30.620683],[104.083797,30.620698] ,[104.081851,30.620675],[104.080276,30.620622],[104.077718,30.620695],[104.076107,30.620588],[104.074552,30.620557],[104.07202,30.620715],[104.071711,30.622295],[104.071833,30.624357],[ 104.072982,30.625105],[104.073218,30.625109],[104.074141,30.625175],[104.076509,30.62515],[104.076778,30.622888],[104.076874,30.621773],[104.076716,30.624068],[104.0766,30.625956],[104.076441, 30.628046],[104.076338,30.630266],[104.076184,30.632202],[104.076129,30.633659],[104.076051,30.635315],[104.075996,30.637499]).

The trajectory data of the driving route R ₁ starting at 12:01 on a certain day, the driving route R ₂ is the trajectory data starting at 18:13 on a certain day, the real time of driving the red track is 3481 seconds, and the real time of driving the blue track is 2550 seconds. Using the method of the present invention, the predicted time after driving the red track is 2975 seconds, and the predicted time after driving the blue track is 2861 seconds. Considering the complexity, uncertainty, traffic lights and other factors of road conditions, and referring to the current overall technical level of the industry, the accuracy of this result is acceptable.

Experiments show that the method according to the present invention can effectively predict the travel time of a given route. The present invention takes into account the influence of real-time road conditions on travel time, and uses a deep learning method to construct a prediction model, thereby obtaining a higher-precision prediction result.

Preferably, the present invention also provides an urban traffic travel time prediction system, which can improve the accuracy of urban traffic travel time prediction.

As shown in FIG. 4 , the urban traffic travel time prediction system of the present invention includes a grid division unit 1, a normalization processing unit 2, a first training unit 3, a simplified processing unit 4, a feature determination unit 5, and a second training unit 6 and temporal prediction unit 7.

Specifically, the grid dividing unit 1 is used to divide the city to be tested into rectangular grids;

The normalization processing unit 2 is configured to construct a normalized floating-point traffic flow matrix according to the historical traffic flow data based on the rectangular grid;

The first training unit 3 is used for training the urban traffic flow prediction network according to the normalized floating-point traffic flow matrix;

The simplification processing unit 4 is configured to perform simplified processing on the trajectory point sequence of the vehicle driving path according to the rectangular grid to obtain the gridded trajectory of the vehicle driving path;

The feature determination unit 5 is configured to determine the feature vector of each track point in the gridded track according to the normalized floating-point traffic flow matrix and the gridded track of the vehicle travel path;

The second training unit 6 is used to train the urban traffic travel time prediction network according to the feature vector of each trajectory point in the gridded trajectory;

The time prediction unit 7 is configured to determine, based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the traveling path of the vehicle to be tested, that the vehicle to be tested has traveled the path to be tested. required travel time.

In addition, the present invention also provides an urban traffic travel time prediction system, including:

processor; and

memory arranged to store computer-executable instructions which, when executed, cause the processor to:

Divide the city to be tested into a rectangular grid;

Based on the rectangular grid, construct a normalized floating-point traffic flow matrix according to historical traffic flow data;

According to the normalized floating-point traffic flow matrix, train the urban traffic flow prediction network;

Simplify the trajectory point sequence of the vehicle's driving path according to the rectangular grid, and obtain the gridded trajectory of the vehicle's driving path;

According to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the eigenvectors of each trajectory point in the gridded trajectory;

According to the feature vector of each trajectory point in the gridded trajectory, train the urban traffic travel time prediction network;

Based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the travel path of the vehicle to be tested, the travel time required for the vehicle to be tested to complete the path to be tested is determined.

In addition, the present invention also provides a computer-readable storage medium storing one or more programs that, when executed by an electronic device including a plurality of application programs, make the one or more programs The electronic device performs the following operations:

Divide the city to be tested into a rectangular grid;

Based on the rectangular grid, construct a normalized floating-point traffic flow matrix according to historical traffic flow data;

According to the normalized floating-point traffic flow matrix, train the urban traffic flow prediction network;

Simplify the trajectory point sequence of the vehicle's driving path according to the rectangular grid, and obtain the gridded trajectory of the vehicle's driving path;

According to the normalized floating-point traffic flow matrix and the gridded trajectory of the vehicle travel path, determine the eigenvectors of each trajectory point in the gridded trajectory;

According to the feature vector of each trajectory point in the gridded trajectory, train the urban traffic travel time prediction network;

Based on the urban traffic flow prediction network and the urban travel time prediction network, and according to the travel path of the vehicle to be tested, the travel time required for the vehicle to be tested to complete the path to be tested is determined.

Compared with the prior art, the urban traffic travel time prediction system and the computer-readable storage medium of the present invention have the same beneficial effects as the above-mentioned urban traffic travel time prediction method, which will not be repeated here.

So far, the technical solutions of the present invention have been described with reference to the preferred embodiments shown in the accompanying drawings, however, those skilled in the art can easily understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims (5)

1. A prediction method for urban traffic travel time is characterized by comprising the following steps:

dividing a city to be tested into rectangular grids;

based on the rectangular grid, a normalized floating point traffic flow matrix is constructed according to historical traffic flow data;

training an urban traffic flow prediction network according to the normalized floating point traffic flow matrix; the method specifically comprises the following steps:

according to the normalized floating point traffic flow matrix, constructing a corresponding first training sample:

P_t＝(A_t；B_t)，

wherein t represents a time interval number, P_tRepresenting a first training sample; a. the_tRepresenting a first training sample P_tThe sample characteristic is a three-dimensional tensor of L rows, M columns and J layers, and a normalized traffic flow matrix obtained in the first J continuous time intervals with the t-th time interval as an end point is piled in a time sequenceAnd (3) stacking: a. the_t＝[Y_t-J+1,Y_t-J+2,…,Y_t-1,Y_t]；

B_tRepresenting a first training sample P_tThe sample mark is a three-dimensional tensor of an L-row M-column Q layer, and is formed by stacking normalized traffic flow matrixes obtained in Q continuous time intervals with the t +1 th time interval as a starting point according to a time sequence: b is_t＝[Y_t+1,Y_t+2,…,Y_t+Q]；

Wherein, Y_t-J+1Is a normalized floating point traffic flow matrix, Y, obtained in the t-J +1 time interval_tFor the normalized floating-point traffic flow matrix, Y, obtained in the t-th time interval_t+QObtaining a normalized floating point traffic flow matrix in the t + Q time intervals;

training an urban traffic flow prediction network according to the first training sample; the urban traffic flow prediction network is used for predicting the normalized floating point traffic flow of a rectangular grid G in Q continuous time intervals;

simplifying the track point sequence of the vehicle running path according to the rectangular grid to obtain a gridded track of the vehicle running path; the method specifically comprises the following steps:

each vehicle driving path is described by a track point sequence, and each track point comprises a longitude coordinate and a latitude coordinate of the track point;

for a given vehicle travel path, the corresponding trajectory sequence is T, which is described by N trajectory points: t { (x)₁,y₁),(x₂,y₂),…,(x_N,y_N)}；

Wherein x is₁And y₁Respectively representing the longitude and latitude, x, of the first trace point_NAnd y_NRespectively representing the longitude and the latitude of the Nth track point;

according to the rectangular grid G, dividing N track points of the track sequence T into different segments in sequence, and enabling the track points in each segment to be located in the same grid area of the rectangular grid G:

wherein, let the trajectory T be divided into n segments, the first segment containing the point (x) from the first trajectory₁,y₁) To d₁One track point

The second fragment comprises the second fragment from₁+1 track points

To d₂One track point

By analogy, the nth segment contains the segment from d_n-1+1 track points

To the Nth track point (x)_N,y_N) And | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number;

calculating the average value of the longitude and the latitude of the track points in the same segment to obtain a gridded track T_G：

Wherein,

represents the average of the longitudes of track points belonging to the ith segment,

mean value, x, representing the latitude of a tracing point belonging to the ith segment_jAnd y_jAre respectively asLongitude and latitude coordinates, when i is 1, d₀1 is ═ 1; when i is n, d_n＝N；

Determining the characteristic vector of each track point in the gridding track according to the normalized floating point traffic flow matrix and the gridding track of the vehicle running path; the method specifically comprises the following steps:

taking out continuous Q normalized floating point traffic flow matrixes Y from the normalized floating point traffic flow matrix_p+1,Y_p+2,…,Y_p+Q(ii) a Wherein p is the first track point (x) of the vehicle driving path T₁,y₁) The sequence number of the time interval to which the time belongs in the training time interval;

according to the gridding track T_GThe ith track point of

R is located at rectangular grid G_iRow and c_iTaking out the normalized floating point traffic flow matrix Y from the grid area where the row is located_p+1,Y_p+2,…,Y_p+QMiddle r_iRow and c_iElements of the column:

wherein,

for normalizing floating-point traffic flow matrix Y_p+QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded track T_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids;

constructing a gridded trajectory T_GCharacteristic vector f of ith track point_i：

Wherein n is a gridding track T_GThe number of the track points; superscript T represents vector transposition;

training a prediction network of urban traffic travel time according to the feature vectors of all track points in the gridding track; the method specifically comprises the following steps:

acquiring the recording time o of each track point in the track sequence T of the vehicle running path₁,o₂,…,o_N(ii) a Wherein o is_NRecording time of the Nth track point;

according to the rectangular grid G, dividing each recording time into different segments in sequence, and enabling the recording time in each segment to be located in the same grid area of the rectangular grid G:

wherein it is provided that the recording time is divided into n segments, the first segment containing the time from the first recording time o₁To d₁A recording time

The second fragment comprises the second fragment from₁+1 recording time

To d₂A recording time

By analogy, the nth segment contains the segment from d_n-1+1 recording time

To the Nth recording time o_NAnd | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number | | | represents a fragment space character, d₁、d₂、d_n-1N is a natural number;

calculating the driving time required by driving the current section i:

wherein,

for the time of the i-th segment, corresponding to the feature vector f_iThe marking time of (1); i. j, d_i、d_i-1Are all natural numbers, and when i is equal to 1, d₀1 is ═ 1; when i is n, d_n+1＝N；

And constructing a second training sample according to the feature vector and the driving time of each segment after driving:

where F represents the second training sample, the sequence of feature vectors F₁,f₂…,f_nFor the sample characteristics of the second training sample F, time sequence

Labeling the sample of the second training sample F; f. of_nThe characteristic vector of the nth track point of the gridding track of the vehicle running path is obtained;

to correspond to the feature vector f_nThe marking time of (1);

training an urban traffic travel time prediction network according to the second training sample, wherein the urban traffic travel time prediction network is used for predicting the travel time of the whole travel path;

determining the travel time required by the vehicle to be tested to finish running the path to be tested according to the running path to be tested of the vehicle to be tested on the basis of the urban traffic flow prediction network and the urban travel time prediction network; the method specifically comprises the following steps:

calculating normalized floating point traffic flow matrixes in the first J continuous time intervals by taking the current running time of the vehicle to be detected as a terminal time interval, and stacking the matrixes into a three-dimensional tensor according to a time sequence:

A＝[Y₁,Y₂,…,Y_J]；

a is a three-dimensional tensor formed by three-dimensional stereo data of L rows, M columns and J layers, and is formed by stacking normalized floating point traffic flow matrixes obtained in the first J continuous time intervals with the current time interval as an end point according to a time sequence, and Y is₁For the normalized floating point traffic flow matrix, Y, obtained in the first J-1 time interval₂For the normalized floating point traffic flow matrix, Y, obtained in the first J-2 time intervals_JObtaining a normalized floating point traffic flow matrix in the current time; l is the number of rows of the rectangular grid G in the latitude direction, and M is the number of columns of the rectangular grid G in the longitude direction;

according to the three-dimensional tensor A and an urban traffic flow prediction network, predicting normalized floating point traffic flow in Q time intervals in future: b is₁,B₂,…,B_Q；

Wherein, B_QPredicting the normalized floating point traffic flow in the future Q-th time interval predicted by the network for the urban traffic flow;

determining a corresponding gridding track R according to a track sequence R of a running path to be tested of a vehicle to be tested_GFeature vectors of the track points;

sample feature B of trace sequence R: b ═ B₁,b₂,…,b_n)；

Wherein, the sample characteristic B of the running path R to be measured is composed of a characteristic vector sequence B₁,b₂,…,b_nComposition is carried out; b₁Is a gridded track R_GThe feature vector of the first trace point of (b)_nIs a gridded track R_GThe feature vector of the nth trace point; n is a gridded locus R_GThe number of the contained track points;

gridding track R of vehicle running path R to be tested_GThe feature vector of the ith trace point of (1):

wherein, b_iRepresenting a gridded footprint R_GThe feature vector of the ith trace point of (a),

and

respectively represent gridded traces R_GLongitude and latitude of the ith trace point,

floating point normalized traffic flow moment B in the first time interval predicted by urban traffic flow prediction network at present moment₁R of_iRow and c_iThe elements of the column are,

a floating point normalized traffic flow matrix B in a second time interval predicted by the urban traffic flow prediction network at the current moment₂R of_iRow and c_iThe elements of the column are,

floating point normalization traffic flow matrix B in Q time interval predicted by urban traffic flow prediction network at present moment_QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded trace R_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids; the natural number Q is the number of time intervals predicted by the urban traffic flow prediction network, and the superscript T represents vector transposition;

from said-gridded footprint R_GAnd determining the time of the running path R to be tested of the vehicle to be tested after the vehicle to be tested runs.

2. The method for predicting urban traffic travel time according to claim 1, wherein the constructing a normalized floating-point traffic flow matrix according to historical traffic flow data based on the rectangular grid specifically comprises:

will select history n_dDays are training periods, each day of which is divided into K with fixed time intervals_dA time interval; the total number of consecutive time intervals into which the training period is divided is K: k is n_d×K_d；

Counting the track data of the vehicles entering the networking system in the rectangular grid G in the training period, wherein the longitude position, the latitude position and the time information of the vehicles are recorded in each track data;

representing the number of vehicles appearing in the ith row and jth column grid sub-area of the rectangular grid G in the tth time interval;

traversing all time intervals in the training period, counting the vehicle flow entering the networking system in the same time interval to obtain K traffic flow matrixes which are respectively marked as X₁,X₂,...,X_k,...,X_K；X_kA traffic flow matrix in a K-th time interval, K being 1, 2.... K;

according to the following formula, carrying out normalization floating point processing on each traffic flow matrix to obtain a corresponding normalization floating point traffic flow matrix:

wherein,

for the normalized floating point traffic flow matrix Y in the t-th time interval_tRow i and column j elements of (1), x₀Represents the maximum of all elements in the K traffic flow matrices.

3. A system for predicting urban traffic travel time, the system comprising:

the grid dividing unit is used for dividing the city to be detected into rectangular grids;

the normalization processing unit is used for constructing a normalization floating point traffic flow matrix according to historical traffic flow data based on the rectangular grid;

the first training unit is used for training an urban traffic flow prediction network according to the normalized floating point traffic flow matrix; the method specifically comprises the following steps:

according to the normalized floating point traffic flow matrix, constructing a corresponding first training sample:

P_t＝(A_t；B_t)，

wherein t represents a time interval number, P_tRepresenting a first training sample; a. the_tRepresenting a first training sample P_tThe sample characteristic is a three-dimensional tensor of L rows, M columns and J layers, and is formed by stacking normalized traffic flow matrixes obtained in the first J continuous time intervals with the t-th time interval as an end point according to the time sequence: a. the_t＝[Y_t-J+1,Y_t-J+2,…,Y_t-1,Y_t]；

B_tRepresenting a first training sample P_tIs a three-dimensional tensor of L rows, M columns and Q layers, and is marked by the t +1 thThe normalized traffic flow matrixes obtained in Q continuous time intervals with the time intervals as the starting points are stacked according to the time sequence: b is_t＝[Y_t+1,Y_t+2,…,Y_t+Q]；

Wherein, Y_t-J+1Is a normalized floating point traffic flow matrix, Y, obtained in the t-J +1 time interval_tFor the normalized floating-point traffic flow matrix, Y, obtained in the t-th time interval_t+QObtaining a normalized floating point traffic flow matrix in the t + Q time intervals;

training an urban traffic flow prediction network according to the first training sample; the urban traffic flow prediction network is used for predicting the normalized floating point traffic flow of a rectangular grid G in Q continuous time intervals;

the simplification processing unit is used for simplifying the track point sequence of the vehicle driving path according to the rectangular grid to obtain a gridded track of the vehicle driving path; the method specifically comprises the following steps:

each vehicle driving path is described by a track point sequence, and each track point comprises a longitude coordinate and a latitude coordinate of the track point;

for a given vehicle travel path, the corresponding trajectory sequence is T, which is described by N trajectory points: t { (x)₁,y₁),(x₂,y₂),…,(x_N,y_N)}；

Wherein x is₁And y₁Respectively representing the longitude and latitude, x, of the first trace point_NAnd y_NRespectively representing the longitude and the latitude of the Nth track point;

according to the rectangular grid G, dividing N track points of the track sequence T into different segments in sequence, and enabling the track points in each segment to be located in the same grid area of the rectangular grid G:

wherein it is assumed that the trajectory T will be divided into n segments, the firstA segment contains a point (x) from the first track₁,y₁) To d₁One track point

The second fragment comprises the second fragment from₁+1 track points

To d₂One track point

By analogy, the nth segment contains the segment from d_n-1+1 track points

To the Nth track point (x)_N,y_N) And | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number;

calculating the average value of the longitude and the latitude of the track points in the same segment to obtain a gridded track T_G：

Wherein,

represents the average of the longitudes of track points belonging to the ith segment,

mean value, x, representing the latitude of a tracing point belonging to the ith segment_jAnd y_jLongitude coordinate and latitude coordinate, when i is 1, d₀1 is ═ 1; when i is n, d_n＝N；

The characteristic determining unit is used for determining the characteristic vector of each track point in the gridding track according to the normalized floating point traffic flow matrix and the gridding track of the vehicle running path; the method specifically comprises the following steps:

taking out continuous Q normalized floating point traffic flow matrixes Y from the normalized floating point traffic flow matrix_p+1,Y_p+2,…,Y_p+Q(ii) a Wherein p is the first track point (x) of the vehicle driving path T₁,y₁) The sequence number of the time interval to which the time belongs in the training time interval;

according to the gridding track T_GThe ith track point of

R is located at rectangular grid G_iRow and c_iTaking out the normalized floating point traffic flow matrix Y from the grid area where the row is located_p+1,Y_p+2,…,Y_p+QMiddle r_iRow and c_iElements of the column:

wherein,

for normalizing floating-point traffic flow matrix Y_p+QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded track T_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids;

constructing a gridded trajectory T_GCharacteristic vector f of ith track point_i：

Wherein n is a gridding track T_GThe number of the track points; superscript T represents vector transposition;

the second training unit is used for training the urban traffic travel time prediction network according to the feature vectors of all track points in the gridding track; the method specifically comprises the following steps:

acquiring the recording time o of each track point in the track sequence T of the vehicle running path₁,o₂,…,o_N(ii) a Wherein o is_NRecording time of the Nth track point;

according to the rectangular grid G, dividing each recording time into different segments in sequence, and enabling the recording time in each segment to be located in the same grid area of the rectangular grid G:

wherein it is provided that the recording time is divided into n segments, the first segment containing the time from the first recording time o₁To d₁A recording time

The second fragment comprises the second fragment from₁+1 recording time

To d₂A recording time

By analogy, the nth segment contains the segment from d_n-1+1 recording time

To the Nth recording time o_NAnd | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number | | | represents a fragment space character, d₁、d₂、d_n-1N is naturalCounting;

calculating the driving time required by driving the current section i:

wherein,

for the time of the i-th segment, corresponding to the feature vector f_iThe marking time of (1); i. j, d_i、d_i-1Are all natural numbers, and when i is equal to 1, d₀1 is ═ 1; when i is n, d_n+1＝N；

And constructing a second training sample according to the feature vector and the driving time of each segment after driving:

where F represents the second training sample, the sequence of feature vectors F₁,f₂…,f_nFor the sample characteristics of the second training sample F, time sequence

Labeling the sample of the second training sample F; f. of_nThe characteristic vector of the nth track point of the gridding track of the vehicle running path is obtained;

to correspond to the feature vector f_nThe marking time of (1);

training an urban traffic travel time prediction network according to the second training sample, wherein the urban traffic travel time prediction network is used for predicting the travel time of the whole travel path;

the time prediction unit is used for determining the travel time required by the vehicle to be detected to run through the path to be detected according to the to-be-detected running path of the vehicle to be detected based on the urban traffic flow prediction network and the urban travel time prediction network; the method specifically comprises the following steps:

calculating normalized floating point traffic flow matrixes in the first J continuous time intervals by taking the current running time of the vehicle to be detected as a terminal time interval, and stacking the matrixes into a three-dimensional tensor according to a time sequence:

A＝[Y₁,Y₂,…,Y_J]；

a is a three-dimensional tensor formed by three-dimensional stereo data of L rows, M columns and J layers, and is formed by stacking normalized floating point traffic flow matrixes obtained in the first J continuous time intervals with the current time interval as an end point according to a time sequence, and Y is₁For the normalized floating point traffic flow matrix, Y, obtained in the first J-1 time interval₂For the normalized floating point traffic flow matrix, Y, obtained in the first J-2 time intervals_JObtaining a normalized floating point traffic flow matrix in the current time; l is the number of rows of the rectangular grid G in the latitude direction, and M is the number of columns of the rectangular grid G in the longitude direction;

according to the three-dimensional tensor A and an urban traffic flow prediction network, predicting normalized floating point traffic flow in Q time intervals in future: b is₁,B₂,…,B_Q；

Wherein, B_QPredicting the normalized floating point traffic flow in the future Q-th time interval predicted by the network for the urban traffic flow;

determining a corresponding gridding track R according to a track sequence R of a running path to be tested of a vehicle to be tested_GFeature vectors of the track points;

sample feature B of trace sequence R: b ═ B₁,b₂,…,b_n)；

Wherein, the sample characteristic B of the running path R to be measured is composed of a characteristic vector sequence B₁,b₂,…,b_nComposition is carried out; b₁Is a gridded track R_GThe feature vector of the first trace point of (b)_nIs a gridded track R_GThe feature vector of the nth trace point; n is a gridded locus R_GThe number of the contained track points;

gridding track R of vehicle running path R to be tested_GThe feature vector of the ith trace point of (1):

wherein, b_iRepresenting a gridded footprint R_GThe feature vector of the ith trace point of (a),

and

respectively represent gridded traces R_GLongitude and latitude of the ith trace point,

floating point normalized traffic flow moment B in the first time interval predicted by urban traffic flow prediction network at present moment₁R of_iRow and c_iThe elements of the column are,

a floating point normalized traffic flow matrix B in a second time interval predicted by the urban traffic flow prediction network at the current moment₂R of_iRow and c_iThe elements of the column are,

floating point normalization traffic flow matrix B in Q time interval predicted by urban traffic flow prediction network at present moment_QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded trace R_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids; the natural number Q is the number of time intervals predicted by the urban traffic flow prediction network, and the superscript T represents vector transposition;

from said-gridded footprint R_GAnd determining the time of the running path R to be tested of the vehicle to be tested after the vehicle to be tested runs.

4. A system for predicting urban traffic travel time, comprising:

a processor; and

a memory arranged to store computer executable instructions that, when executed, cause the processor to:

dividing a city to be tested into rectangular grids;

based on the rectangular grid, a normalized floating point traffic flow matrix is constructed according to historical traffic flow data;

training an urban traffic flow prediction network according to the normalized floating point traffic flow matrix; the method specifically comprises the following steps:

according to the normalized floating point traffic flow matrix, constructing a corresponding first training sample:

P_t＝(A_t；B_t)，

wherein t represents a time interval number, P_tRepresenting a first training sample; a. the_tRepresenting a first training sample P_tThe sample characteristic is a three-dimensional tensor of L rows, M columns and J layers, and is formed by stacking normalized traffic flow matrixes obtained in the first J continuous time intervals with the t-th time interval as an end point according to the time sequence: a. the_t＝[Y_t-J+1,Y_t-J+2,…,Y_t-1,Y_t]；

B_tRepresenting a first training sample P_tThe sample mark is a three-dimensional tensor of an L-row M-column Q layer, and is formed by stacking normalized traffic flow matrixes obtained in Q continuous time intervals with the t +1 th time interval as a starting point according to a time sequence: b is_t＝[Y_t+1,Y_t+2,…,Y_t+Q]；

Wherein, Y_t-J+1Is a normalized floating point traffic flow matrix, Y, obtained in the t-J +1 time interval_tFor the normalized floating-point traffic flow matrix, Y, obtained in the t-th time interval_t+QObtaining a normalized floating point traffic flow matrix in the t + Q time intervals;

training an urban traffic flow prediction network according to the first training sample; the urban traffic flow prediction network is used for predicting the normalized floating point traffic flow of a rectangular grid G in Q continuous time intervals;

simplifying the track point sequence of the vehicle running path according to the rectangular grid to obtain a gridded track of the vehicle running path; the method specifically comprises the following steps:

each vehicle driving path is described by a track point sequence, and each track point comprises a longitude coordinate and a latitude coordinate of the track point;

for a given vehicle travel path, the corresponding trajectory sequence is T, which is described by N trajectory points: t { (x)₁,y₁),(x₂,y₂),…,(x_N,y_N)}；

Wherein x is₁And y₁Respectively representing the longitude and latitude, x, of the first trace point_NAnd y_NRespectively representing the longitude and the latitude of the Nth track point;

according to the rectangular grid G, dividing N track points of the track sequence T into different segments in sequence, and enabling the track points in each segment to be located in the same grid area of the rectangular grid G:

wherein, let the trajectory T be divided into n segments, the first segment containing the point (x) from the first trajectory₁,y₁) To d₁One track point

The second fragment comprises the second fragment from₁+1 track points

To d₂One track point

By analogy, the nth segment contains the segment from d_n-1+1 track points

To the Nth track point (x)_N,y_N) And | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number;

calculating the average value of the longitude and the latitude of the track points in the same segment to obtain a gridded track T_G：

Wherein,

represents the average of the longitudes of track points belonging to the ith segment,

mean value, x, representing the latitude of a tracing point belonging to the ith segment_jAnd y_jLongitude coordinate and latitude coordinate, when i is 1, d₀1 is ═ 1; when i is n, d_n＝N；

Determining the characteristic vector of each track point in the gridding track according to the normalized floating point traffic flow matrix and the gridding track of the vehicle running path; the method specifically comprises the following steps:

taking out continuous Q normalized floating point traffic flow matrixes Y from the normalized floating point traffic flow matrix_p+1,Y_p+2,…,Y_p+Q(ii) a Wherein p is the first track point (x) of the vehicle driving path T₁,y₁) The sequence number of the time interval to which the time belongs in the training time interval;

according to the gridding track T_GThe ith track point of

R is located at rectangular grid G_iRow and c_iTaking out the normalized floating point traffic flow matrix Y from the grid area where the row is located_p+1,Y_p+2,…,Y_p+QMiddle r_iRow and c_iElements of the column:

wherein,

for normalizing floating-point traffic flow matrix Y_p+QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded track T_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids;

constructing a gridded trajectory T_GCharacteristic vector f of ith track point_i：

Wherein n is a gridding track T_GThe number of the track points; superscript T represents vector transposition;

training a prediction network of urban traffic travel time according to the feature vectors of all track points in the gridding track; the method specifically comprises the following steps:

acquiring the recording time o of each track point in the track sequence T of the vehicle running path₁,o₂,…,o_N(ii) a Wherein o is_NRecording time of the Nth track point;

according to the rectangular grid G, dividing each recording time into different segments in sequence, and enabling the recording time in each segment to be located in the same grid area of the rectangular grid G:

wherein it is provided that the recording time is divided into n segments, the first segment containing the time from the first recording time o₁To d₁A recording time

The second fragment comprises the second fragment from₁+1 recording time

To d₂A recording time

By analogy, the nth segment contains the segment from d_n-1+1 recording time

To the Nth recording time o_NAnd | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number | | | represents a fragment space character, d₁、d₂、d_n-1N is a natural number;

calculating the driving time required by driving the current section i:

wherein,

for the time of the i-th segment, corresponding to the feature vector f_iThe marking time of (1); i. j, d_i、d_i-1Are all natural numbers, and when i is equal to 1, d₀1 is ═ 1; when i is n, d_n+1＝N；

And constructing a second training sample according to the feature vector and the driving time of each segment after driving:

where F represents the second training sample, the sequence of feature vectors F₁,f₂…,f_nFor the sample characteristics of the second training sample F, time sequence

Labeling the sample of the second training sample F; f. of_nThe characteristic vector of the nth track point of the gridding track of the vehicle running path is obtained;

to correspond to the feature vector f_nThe marking time of (1);

training an urban traffic travel time prediction network according to the second training sample, wherein the urban traffic travel time prediction network is used for predicting the travel time of the whole travel path;

determining the travel time required by the vehicle to be tested to finish running the path to be tested according to the running path to be tested of the vehicle to be tested on the basis of the urban traffic flow prediction network and the urban travel time prediction network; the method specifically comprises the following steps:

calculating normalized floating point traffic flow matrixes in the first J continuous time intervals by taking the current running time of the vehicle to be detected as a terminal time interval, and stacking the matrixes into a three-dimensional tensor according to a time sequence:

A＝[Y₁,Y₂,…,Y_J]；

a is a three-dimensional tensor formed by three-dimensional stereo data of L rows, M columns and J layers, and is formed by stacking normalized floating point traffic flow matrixes obtained in the first J continuous time intervals with the current time interval as an end point according to a time sequence, and Y is₁For the normalized floating point traffic flow matrix, Y, obtained in the first J-1 time interval₂For the normalized floating point traffic flow matrix, Y, obtained in the first J-2 time intervals_JObtaining a normalized floating point traffic flow matrix in the current time; l is the number of rows of the rectangular grid G in the latitude direction, and M is the number of columns of the rectangular grid G in the longitude direction;

according to the three-dimensional tensor A and an urban traffic flow prediction network, predicting normalized floating point traffic flow in Q time intervals in future: b is₁,B₂,…,B_Q；

Wherein, B_QPredicting the normalized floating point traffic flow in the future Q-th time interval predicted by the network for the urban traffic flow;

determining a corresponding gridding track R according to a track sequence R of a running path to be tested of a vehicle to be tested_GFeature vectors of the track points;

sample feature B of trace sequence R: b ═ B₁,b₂,…,b_n)；

Wherein, the sample characteristic B of the running path R to be measured is composed of a characteristic vector sequence B₁,b₂,…,b_nComposition is carried out; b₁Is a gridded track R_GThe feature vector of the first trace point of (b)_nIs a gridded track R_GThe feature vector of the nth trace point; n is a gridded locus R_GThe number of the contained track points;

gridding track R of vehicle running path R to be tested_GThe feature vector of the ith trace point of (1):

wherein, b_iRepresenting a gridded footprint R_GThe feature vector of the ith trace point of (a),

and

respectively represent gridded traces R_GLongitude and latitude of the ith trace point,

floating point normalized traffic flow moment B in the first time interval predicted by urban traffic flow prediction network at present moment₁R of_iRow and c_iThe elements of the column are,

a floating point normalized traffic flow matrix B in a second time interval predicted by the urban traffic flow prediction network at the current moment₂R of_iRow and c_iThe elements of the column are,

floating point normalization traffic flow matrix B in Q time interval predicted by urban traffic flow prediction network at present moment_QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded trace R_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids; the natural number Q is the number of time intervals predicted by the urban traffic flow prediction network, and the superscript T represents vector transposition;

from said-gridded footprint R_GAnd determining the time of the running path R to be tested of the vehicle to be tested after the vehicle to be tested runs.

5. A computer-readable storage medium storing one or more programs that, when executed by an electronic device including a plurality of application programs, cause the electronic device to:

dividing a city to be tested into rectangular grids;

based on the rectangular grid, a normalized floating point traffic flow matrix is constructed according to historical traffic flow data;

training an urban traffic flow prediction network according to the normalized floating point traffic flow matrix; the method specifically comprises the following steps:

according to the normalized floating point traffic flow matrix, constructing a corresponding first training sample:

P_t＝(A_t；B_t)，

wherein t represents a time interval number, P_tRepresenting a first training sample; a. the_tRepresenting a first training sample P_tThe sample characteristic is a three-dimensional tensor of L rows, M columns and J layers, and is formed by stacking normalized traffic flow matrixes obtained in the first J continuous time intervals with the t-th time interval as an end point according to the time sequence: a. the_t＝[Y_t-J+1,Y_t-J+2,…,Y_t-1,Y_t]；

B_tRepresenting a first training sample P_tThe sample mark is a three-dimensional tensor of an L-row M-column Q layer, and is formed by stacking normalized traffic flow matrixes obtained in Q continuous time intervals with the t +1 th time interval as a starting point according to a time sequence: b is_t＝[Y_t+1,Y_t+2,…,Y_t+Q]；

Wherein, Y_t-J+1Is a normalized floating point traffic flow matrix, Y, obtained in the t-J +1 time interval_tFor normalized floating point obtained in the t-th time intervalTraffic flow matrix, Y_t+QObtaining a normalized floating point traffic flow matrix in the t + Q time intervals;

training an urban traffic flow prediction network according to the first training sample; the urban traffic flow prediction network is used for predicting the normalized floating point traffic flow of a rectangular grid G in Q continuous time intervals;

simplifying the track point sequence of the vehicle running path according to the rectangular grid to obtain a gridded track of the vehicle running path; the method specifically comprises the following steps:

each vehicle driving path is described by a track point sequence, and each track point comprises a longitude coordinate and a latitude coordinate of the track point;

for a given vehicle travel path, the corresponding trajectory sequence is T, which is described by N trajectory points: t { (x)₁,y₁),(x₂,y₂),…,(x_N,y_N)}；

Wherein x is₁And y₁Respectively representing the longitude and latitude, x, of the first trace point_NAnd y_NRespectively representing the longitude and the latitude of the Nth track point;

according to the rectangular grid G, dividing N track points of the track sequence T into different segments in sequence, and enabling the track points in each segment to be located in the same grid area of the rectangular grid G:

wherein, let the trajectory T be divided into n segments, the first segment containing the point (x) from the first trajectory₁,y₁) To d₁One track point

The second fragment comprises the second fragment from₁+1 track points

To d₂One track point

By analogy, the nth segment contains the segment from d_n-1+1 track points

To the Nth track point (x)_N,y_N) And | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number;

calculating the average value of the longitude and the latitude of the track points in the same segment to obtain a gridded track T_G：

Wherein,

represents the average of the longitudes of track points belonging to the ith segment,

mean value, x, representing the latitude of a tracing point belonging to the ith segment_jAnd y_jLongitude coordinate and latitude coordinate, when i is 1, d₀1 is ═ 1; when i is n, d_n＝N；

Determining the characteristic vector of each track point in the gridding track according to the normalized floating point traffic flow matrix and the gridding track of the vehicle running path; the method specifically comprises the following steps:

taking out continuous Q normalized floating point traffic flow matrixes Y from the normalized floating point traffic flow matrix_p+1,Y_p+2,…,Y_p+Q(ii) a Wherein p is a vehicleFirst track point (x) of travel path T₁,y₁) The sequence number of the time interval to which the time belongs in the training time interval;

according to the gridding track T_GThe ith track point of

R is located at rectangular grid G_iRow and c_iTaking out the normalized floating point traffic flow matrix Y from the grid area where the row is located_p+1,Y_p+2,…,Y_p+QMiddle r_iRow and c_iElements of the column:

wherein,

for normalizing floating-point traffic flow matrix Y_p+QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded track T_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids;

constructing a gridded trajectory T_GCharacteristic vector f of ith track point_i：

Wherein n is a gridding track T_GThe number of the track points; superscript T represents vector transposition;

training a prediction network of urban traffic travel time according to the feature vectors of all track points in the gridding track; the method specifically comprises the following steps:

acquiring the recording time o of each track point in the track sequence T of the vehicle running path₁,o₂,…,o_N(ii) a Wherein o is_NRecording time of the Nth track point;

according to the rectangular grid G, dividing each recording time into different segments in sequence, and enabling the recording time in each segment to be located in the same grid area of the rectangular grid G:

wherein it is provided that the recording time is divided into n segments, the first segment containing the time from the first recording time o₁To d₁A recording time

The second fragment comprises the second fragment from₁+1 recording time

To d₂A recording time

By analogy, the nth segment contains the segment from d_n-1+1 recording time

To the Nth recording time o_NAnd | represents a fragment space symbol; d₁、d₂、d_n-1N is a natural number | | | represents a fragment space character, d₁、d₂、d_n-1N is a natural number;

calculating the driving time required by driving the current section i:

wherein,

for the time of the i-th segment, corresponding to the feature vector f_iThe marking time of (1); i. j, d_i、d_i-1Are all natural numbers, and when i is equal to 1, d₀1 is ═ 1; when i is n, d_n+1＝N；

And constructing a second training sample according to the feature vector and the driving time of each segment after driving:

where F represents the second training sample, the sequence of feature vectors F₁,f₂…,f_nFor the sample characteristics of the second training sample F, time sequence

Labeling the sample of the second training sample F; f. of_nThe characteristic vector of the nth track point of the gridding track of the vehicle running path is obtained;

to correspond to the feature vector f_nThe marking time of (1);

training an urban traffic travel time prediction network according to the second training sample, wherein the urban traffic travel time prediction network is used for predicting the travel time of the whole travel path;

determining the travel time required by the vehicle to be tested to finish running the path to be tested according to the running path to be tested of the vehicle to be tested on the basis of the urban traffic flow prediction network and the urban travel time prediction network; the method specifically comprises the following steps:

calculating normalized floating point traffic flow matrixes in the first J continuous time intervals by taking the current running time of the vehicle to be detected as a terminal time interval, and stacking the matrixes into a three-dimensional tensor according to a time sequence:

A＝[Y₁,Y₂,…,Y_J]；

a is a three-dimensional tensor formed by three-dimensional stereo data of L rows, M columns and J layers, and is formed by stacking normalized floating point traffic flow matrixes obtained in the first J continuous time intervals with the current time interval as an end point according to a time sequence, and Y is₁For the normalized floating point traffic flow matrix, Y, obtained in the first J-1 time interval₂For the normalized floating point traffic flow matrix, Y, obtained in the first J-2 time intervals_JObtaining a normalized floating point traffic flow matrix in the current time; l is the number of rows of the rectangular grid G in the latitude direction, and M is the number of columns of the rectangular grid G in the longitude direction;

according to the three-dimensional tensor A and an urban traffic flow prediction network, predicting normalized floating point traffic flow in Q time intervals in future: b is₁,B₂,…,B_Q；

Wherein, B_QPredicting the normalized floating point traffic flow in the future Q-th time interval predicted by the network for the urban traffic flow;

determining a corresponding gridding track R according to a track sequence R of a running path to be tested of a vehicle to be tested_GFeature vectors of the track points;

sample feature B of trace sequence R: b ═ B₁,b₂,…,b_n)；

Wherein, the sample characteristic B of the running path R to be measured is composed of a characteristic vector sequence B₁,b₂,…,b_nComposition is carried out; b₁Is a gridded track R_GThe feature vector of the first trace point of (b)_nIs a gridded track R_GThe feature vector of the nth trace point; n is a gridded locus R_GThe number of the contained track points;

gridding track R of vehicle running path R to be tested_GThe feature vector of the ith trace point of (1):

wherein, b_iRepresenting a gridded footprint R_GThe ith track ofThe feature vector of a point is determined,

and

respectively represent gridded traces R_GLongitude and latitude of the ith trace point,

floating point normalized traffic flow moment B in the first time interval predicted by urban traffic flow prediction network at present moment₁R of_iRow and c_iThe elements of the column are,

a floating point normalized traffic flow matrix B in a second time interval predicted by the urban traffic flow prediction network at the current moment₂R of_iRow and c_iThe elements of the column are,

floating point normalization traffic flow matrix B in Q time interval predicted by urban traffic flow prediction network at present moment_QR of_iRow and c_iElements of a column; r is_iAnd c_iRespectively, a gridded trace R_GThe ith track point of

The row serial numbers and the column serial numbers of the rectangular grids; the natural number Q is the number of time intervals predicted by the urban traffic flow prediction network, and the superscript T represents vector transposition;

from said-gridded footprint R_GAnd determining the time of the running path R to be tested of the vehicle to be tested after the vehicle to be tested runs.

Images