CN116821831A - Intelligent electric power inspection system and method thereof - Google Patents

Abstract

The application relates to the field of intelligent diagnosis, and particularly discloses an intelligent electric power inspection system and an intelligent electric power inspection method. Therefore, the running state of the power equipment can be accurately judged, potential faults can be found early, and the normal running of the power equipment is ensured.

Description

Intelligent power inspection system and method

Technical field

The present application relates to the field of intelligent diagnosis, and more specifically, to an electric power intelligent inspection system and a method thereof.

Background technique

Electric power inspection is a kind of regular inspection, inspection and maintenance of power equipment. Power inspections aim to detect equipment failures, prevent accidents, deal with equipment defects in a timely manner, and ensure the safe and reliable operation of power equipment. Electrical equipment often changes as usage time increases, and inspection cycles are generally fixed and relatively broad, making it impossible to monitor equipment status in a timely and effective manner. In addition, the quality of inspections is affected by the professionalism of the inspection personnel, making it difficult to guarantee the quality of inspections.

Therefore, an optimized intelligent power inspection solution is needed.

Contents of the invention

In order to solve the above technical problems, this application is proposed. Embodiments of the present application provide an electric power intelligent inspection system and a method thereof, which extract features from the temperature values and vibration signals of power equipment when they are running, and fuse the temperature features and vibration features to determine the power equipment. Is the operation normal? In this way, the operating status of the power equipment can be judged more accurately, potential faults can be discovered early, and the normal operation of the power equipment can be ensured.

According to one aspect of this application, an intelligent power inspection system is provided, which includes:

An equipment data acquisition module, used to obtain equipment temperature values at multiple predetermined time points within a predetermined time period and equipment vibration signals within the predetermined time period collected by vibration sensors;

A temperature feature extraction module, configured to arrange the device temperature values at the plurality of predetermined time points into device temperature input vectors according to the time dimension and then pass the multi-scale neighborhood feature extraction module to obtain the device temperature feature vector;

S transformation module, used to perform S transformation on the equipment vibration signal to obtain S transformation time-frequency diagram;

A vibration coding module, used to pass the S-transformed time-frequency diagram through a convolutional neural network model as a filter to obtain the equipment vibration feature vector;

A fusion module, used to fuse the equipment temperature feature vector and the equipment vibration feature vector to obtain a classification feature matrix; and

An operating status result generation module is used to pass the classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate whether the operating status of the power equipment is normal.

In the above-mentioned electric power intelligent inspection system, the temperature feature extraction module includes: a first scale temperature feature extraction unit for inputting the equipment temperature input vector into the first convolution of the multi-scale neighborhood feature extraction module. layer to obtain a first scale temperature feature vector, wherein the first convolution layer has a first one-dimensional convolution kernel of a first length; a second scale temperature feature extraction unit is used to input the device temperature input vector The second convolution layer of the multi-scale neighborhood feature extraction module is used to obtain the second scale temperature feature vector, wherein the second convolution layer has a second one-dimensional convolution kernel of a second length, and the first The length is different from the second length; and, a multi-scale cascade unit is used to cascade the first-scale temperature feature vector and the second-scale temperature feature vector to obtain the device temperature feature vector.

In the above-mentioned electric power intelligent inspection system, the first-scale temperature feature extraction unit is further configured to: use the first convolution layer of the multi-scale neighborhood feature extraction module to calculate the equipment using the following first convolution formula: The temperature input vector is convolutionally encoded to obtain the first scale temperature feature vector; where the first convolution formula is:

Among them, a is the width of the first convolution kernel in the x direction, F(a) is the first convolution kernel parameter vector, G(xa) is the local vector matrix that operates with the first convolution kernel function, and w is the first convolution kernel parameter vector. The size of a convolution kernel, X represents the device temperature input vector, and Cov ₁ (X) is the first scale temperature feature vector.

In the above-mentioned electric power intelligent inspection system, the second-scale feature extraction unit to be detected is further configured to: use the second convolution layer of the multi-scale neighborhood feature extraction module to use the following second convolution formula to The equipment temperature input vector is convolutionally encoded to obtain the second scale temperature feature vector; wherein, the second convolution formula is:

Among them, b is the width of the second convolution kernel in the x direction, F(b) is the second convolution kernel parameter vector, G(xb) is the local vector matrix that operates with the second convolution kernel function, and m is the second convolution kernel parameter vector. The size of the two convolution kernels, X represents the device temperature input vector, and Cov ₂ (X) is the second scale temperature feature vector.

In the above-mentioned intelligent power inspection system, the S transformation module is used to perform S transformation on the equipment vibration signal using the following transformation formula to obtain the S transformation time-frequency diagram; wherein the transformation formula is:

Among them, s(f,τ) represents the S-transform time-frequency diagram, τ is the time shift factor, x(t) represents the vibration signal of the equipment, f represents the frequency, and t represents the time.

In the above-mentioned intelligent power inspection system, the fusion module includes: a sparse feature vector generation unit for sparsely encoding the equipment temperature feature vector and the equipment vibration feature vector to obtain the first sparse feature vector and the third sparse feature vector. Two sparse feature vectors; a first JS divergence calculation unit, used to calculate the first JS divergence of the first sparse feature vector relative to the second sparse feature vector; a second JS divergence calculation unit, used to calculate a second JS divergence of the second sparse feature vector relative to the first sparse feature vector; a normalization processing unit for normalizing the first JS divergence and the second JS divergence processing to obtain the normalized first JS divergence and the normalized second JS divergence; and, a classification feature matrix generation unit configured to obtain the normalized first JS divergence and the normalized second JS divergence. Two JS divergences are used as weights to fuse the first sparse feature vector and the second sparse feature vector to obtain the classification feature matrix.

In the above-mentioned electric power intelligent inspection system, the first JS divergence calculation unit is used to calculate the first sparse feature vector relative to the second sparse feature vector using the following first JS divergence formula. The first JS divergence; wherein, the first JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₁ represents the first JS divergence.

In the above-mentioned electric power intelligent inspection system, the second JS divergence calculation unit is used to calculate the second sparse eigenvector relative to the first sparse eigenvector according to the following second JS divergence formula. The second JS divergence; wherein, the second JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₂ represents the second JS divergence.

In the above-mentioned intelligent power inspection system, the operating status result generation module includes: an expansion unit for expanding the classification feature matrix into a classification feature vector based on row vectors or column vectors; a fully connected coding unit for The classification feature vector is fully connected using the fully connected layer of the classifier to obtain the coded classification feature vector; and a classification result generation unit is used to classify the coded classification feature vector through the Softmax classification of the classifier. function to obtain the classification results.

According to another aspect of this application, an intelligent power inspection method is provided, which includes:

Obtaining device temperature values at multiple predetermined time points within a predetermined time period and equipment vibration signals within the predetermined time period collected by the vibration sensor;

Arrange the equipment temperature values at the multiple predetermined time points into equipment temperature input vectors according to the time dimension and then pass the multi-scale neighborhood feature extraction module to obtain the equipment temperature feature vector;

Perform S-transform on the equipment vibration signal to obtain an S-transform time-frequency diagram;

Pass the S-transformed time-frequency diagram through the convolutional neural network model as a filter to obtain the equipment vibration feature vector;

Fusion of the equipment temperature feature vector and the equipment vibration feature vector to obtain a classification feature matrix; and

The classification feature matrix is passed through a classifier to obtain a classification result, which is used to indicate whether the operating status of the power equipment is normal.

According to yet another aspect of the present application, an electronic device is provided, including: a processor; and a memory, in which computer program instructions are stored, and when executed by the processor, the computer program instructions cause the The processor executes the power intelligent inspection method as described above.

According to yet another aspect of the present application, a computer-readable medium is provided, with computer program instructions stored thereon. When the computer program instructions are run by a processor, the computer program instructions cause the processor to perform the intelligent power inspection as described above. method.

Compared with the existing technology, the intelligent power inspection system and method provided by this application extract features from the temperature values and vibration signals of the power equipment during operation, and fuse the temperature features and vibration features to determine whether the power equipment is running. Whether the electrical equipment is operating normally. In this way, the operating status of the power equipment can be judged more accurately, potential faults can be discovered early, and the normal operation of the power equipment can be ensured.

Description of the drawings

The above and other objects, features and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. They are used to explain the present application together with the embodiments of the present application and do not constitute a limitation of the present application. In the drawings, like reference numbers generally represent like components or steps.

Figure 1 is a system block diagram of an intelligent power inspection system according to an embodiment of the present application.

Figure 2 is an architectural diagram of an intelligent power inspection system according to an embodiment of the present application.

Figure 3 is a block diagram of the temperature feature extraction module in the intelligent power inspection system according to an embodiment of the present application.

Figure 4 is a block diagram of a fusion module in the intelligent power inspection system according to an embodiment of the present application.

Figure 5 is a block diagram of the operating status result generation module in the intelligent power inspection system according to an embodiment of the present application.

Figure 6 is a flow chart of an intelligent power inspection method according to an embodiment of the present application.

Figure 7 is a block diagram of an electronic device according to an embodiment of the present application.

Detailed ways

Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the example embodiments described here.

Application Overview

As mentioned in the above-mentioned background technology, the inspection period of the existing power inspection method is generally fixed and the inspection period is relatively wide. This easily results in the inability to monitor the status of the power equipment in a timely and effective manner, and the quality of the inspection is affected by the professionalism of the inspection personnel. Due to the horizontal influence, the quality of inspection is difficult to guarantee. Therefore, an optimized power inspection solution is expected, which can effectively and timely monitor the operating status of power equipment and ensure the quality of power inspection.

In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also shown that they are close to or even beyond human performance in areas such as image classification, object detection, semantic segmentation, and text translation. The development of deep learning and neural networks provides new solutions and solutions for power inspection.

Specifically, in the technical solution of the present application, first, the device temperature values at multiple predetermined time points within a predetermined time period and the device vibration signals collected by the vibration sensor within the predetermined time period are obtained. It should be understood that the purpose of obtaining the device temperature values at multiple predetermined time points within a predetermined time period and the device vibration signals collected by the vibration sensor within the predetermined time period is to fully understand the working status and changes of the equipment at different times. Equipment temperature and equipment vibration signals are important reference indicators that reflect the operating status of the equipment. By monitoring and analyzing these indicators, the health status or potential faults of the equipment can be judged, and corresponding measures can be taken to ensure the normal working status of the equipment in a timely manner and ensure The security and stability of important infrastructure.

Next, the device temperature values at the multiple predetermined time points are arranged into device temperature input vectors according to the time dimension and then passed through a multi-scale neighborhood feature extraction module to obtain a device temperature feature vector. The purpose of arranging the equipment temperature values at multiple predetermined time points into equipment temperature input vectors according to the time dimension is to make full use of the temperature change information of the equipment at different time points, thereby more comprehensively reflecting changes in equipment status. The temperature input vector is processed through the multi-scale domain feature extraction module to obtain feature information at different scales. In actual operation, changes in equipment temperature generally have characteristics such as periodicity and time scale differences. Through multi-scale domain feature extraction, it is possible to Provide more comprehensive temperature characteristics, including short-term changes, periodic changes, and duration changes of the equipment, which can more accurately describe the status changes of the equipment and detect potential faults. At the same time, this method can also effectively reduce the impact of noise and improve monitoring accuracy and stability, thereby achieving better intelligent inspection results.

Then, S-transform is performed on the equipment vibration signal to obtain an S-transform time-frequency diagram. It should be understood that since the vibration signal is an unsteady signal with time as the independent variable, it is often difficult to obtain comprehensive information through direct time domain analysis. However, by converting the time domain signal into the time-frequency domain through S transformation, it can simultaneously reflect The signal changes in time and frequency. Thus, while retaining the information of the original signal, the time-frequency characteristics of the vibration signal become more intuitive and clear, which is more conducive to subsequent diagnosis and processing.

Furthermore, the S-transformed time-frequency diagram is passed through the convolutional neural network model as a filter to obtain the equipment vibration feature vector. Those of ordinary skill in the art know that convolutional neural networks perform well in feature extraction. The S-transformed time-frequency diagram contains a large amount of information. High-dimensional data such as the time-frequency diagram can be processed through the convolutional neural network, and features related to the equipment vibration signal can be extracted from it.

Next, the device temperature feature vector and the device vibration feature vector are fused to obtain a classification feature matrix. It should be understood that fusing the equipment temperature feature vector and the equipment vibration feature vector can comprehensively utilize different types of feature information to fully describe multi-directional changes in equipment status. In actual operation, the working status of equipment is usually affected by a variety of factors, such as temperature, vibration and other parameters, among which there is a certain correlation and interdependence between different parameters. The fusion of two key characteristic information, temperature and vibration, can further improve the accuracy and precision of equipment fault diagnosis, thereby better completing intelligent inspections. Finally, the classification matrix is passed through a classifier to obtain an operating result indicating whether the operating status of the power equipment is normal. In this way, the operating status of the equipment can be judged more accurately, avoiding the risk of human misjudgment, while also improving the efficiency of inspections.

It should be understood that, considering that the dimensions and number of samples of the device temperature feature vector and the device vibration feature vector are different, there may be some repeated information or some information irrelevant to the target task in the process of fusing the two feature vectors. , which leads to the problem of noise and redundant information during the fusion process of the equipment temperature feature vector and the equipment vibration feature vector. This information may have an adverse impact on the training of the model and affect the generalization ability of the model.

Based on this, in the technical solution of the present application, fusing the device temperature feature vector and the device vibration feature vector to obtain the classification feature matrix includes: performing the following steps on the device temperature feature vector and the device vibration feature vector: Sparse encoding to obtain the first sparse feature vector and the second sparse feature vector; calculate the first JS divergence of the first sparse feature vector relative to the second sparse feature vector; calculate the second sparse feature vector relative to The second JS divergence of the first sparse feature vector; perform normalization processing on the first JS divergence and the second JS divergence to obtain the normalized first JS divergence and the normalized second JS divergence. Two JS divergences; using the normalized first JS divergence and the normalized second JS divergence as weights, fuse the first sparse feature vector and the second sparse feature vector to obtain the Classification feature matrix.

Specifically, in the technical solution of this application, the first JS divergence of the first sparse eigenvector relative to the second sparse eigenvector is calculated using the following first JS divergence formula; wherein, the first sparse eigenvector is A JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₁ represents the first JS divergence.

Specifically, in the technical solution of the present application, the second JS divergence of the second sparse eigenvector relative to the first sparse eigenvector is calculated using the following second JS divergence formula; wherein, the second sparse eigenvector is The second JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₂ represents the second JS divergence.

The above feature distribution fusion algorithm uses the idea of sparse coding to effectively capture the structure and pattern information between two feature distributions without being affected by noise and redundant information, thereby improving the feature fusion effect. In this way, it can effectively reduce The information loss in the feature fusion process retains the important information in the original feature distribution, improves the quality and credibility of the data after feature fusion, and at the same time, can also effectively reduce the data dimension after feature fusion, reduce data redundancy and noise, and Effectively enhance the data expression ability after feature fusion, extract more implicit information and potential patterns, and improve the ability and level of data mining and knowledge discovery. Thus, the accuracy of the classification result obtained by the classification feature matrix through the classifier is improved.

After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be specifically introduced below with reference to the accompanying drawings.

Example system

Figure 1 is a system block diagram of an intelligent power inspection system according to an embodiment of the present application. As shown in Figure 1, the power intelligent inspection system 100 includes: an equipment data acquisition module 110, which is used to obtain equipment temperature values at multiple predetermined time points within a predetermined time period and the predetermined time period collected by a vibration sensor. The equipment vibration signal within; the temperature feature extraction module 120 is used to arrange the equipment temperature values of the multiple predetermined time points into equipment temperature input vectors according to the time dimension and then pass the multi-scale neighborhood feature extraction module to obtain the equipment temperature feature vector ; S transformation module 130, used to perform S transformation on the device vibration signal to obtain the S transformation time-frequency diagram; Vibration encoding module 140, used to pass the S transformation time-frequency diagram through the convolutional neural network model as a filter to obtain the equipment vibration feature vector; the fusion module 150 is used to fuse the equipment temperature feature vector and the equipment vibration feature vector to obtain the classification feature matrix; and the operating status result generation module 160 is used to fuse the classification feature vector The feature matrix is passed through the classifier to obtain a classification result, which is used to indicate whether the operating status of the power equipment is normal.

Figure 2 is an architectural diagram of an intelligent power inspection system according to an embodiment of the present application. As shown in Figure 2, in this architecture, first, device temperature values at multiple predetermined time points within a predetermined time period and device vibration signals within the predetermined time period collected by vibration sensors are obtained. Next, the device temperature values at the multiple predetermined time points are arranged into device temperature input vectors according to the time dimension and then passed through a multi-scale neighborhood feature extraction module to obtain a device temperature feature vector. Then, S-transform is performed on the equipment vibration signal to obtain an S-transform time-frequency diagram. Furthermore, the S-transformed time-frequency diagram is passed through the convolutional neural network model as a filter to obtain the equipment vibration feature vector. Next, the device temperature feature vector and the device vibration feature vector are fused to obtain a classification feature matrix. Finally, the classification feature matrix is passed through a classifier to obtain a classification result, which is used to indicate whether the operating status of the power equipment is normal.

In the power intelligent inspection system 100, the equipment data acquisition module 110 is used to obtain equipment temperature values at multiple predetermined time points within a predetermined time period and equipment vibration signals within the predetermined time period collected by vibration sensors. It should be understood that the purpose of obtaining the device temperature values at multiple predetermined time points within a predetermined time period and the device vibration signals collected by the vibration sensor within the predetermined time period is to fully understand the working status and changes of the equipment at different times. Equipment temperature and equipment vibration signals are important reference indicators that reflect the operating status of the equipment. By monitoring and analyzing these indicators, the health status or potential faults of the equipment can be judged, and corresponding measures can be taken to ensure the normal working status of the equipment in a timely manner and ensure The security and stability of important infrastructure. Specifically, in the technical solution of the present application, the temperature values at multiple predetermined time points within the predetermined time period can be collected and obtained by temperature sensors, and the vibration signals of the equipment within the predetermined time period are collected and obtained by vibration sensors.

In the power intelligent inspection system 100, the temperature feature extraction module 120 is used to arrange the equipment temperature values of the multiple predetermined time points into equipment temperature input vectors according to the time dimension and then use the multi-scale neighborhood feature extraction module to Get the device temperature feature vector. It should be understood that the purpose of arranging the device temperature values at multiple predetermined time points into the device temperature input vector according to the time dimension is to make full use of the temperature change information of the device at different time points, thereby more comprehensively reflecting the changes in the device status. The multi-scale domain feature extraction module is a commonly used feature extraction method in computer vision. It can extract features at different scales to describe features more comprehensively. The temperature input vector is processed through the multi-scale domain feature extraction module to obtain feature information at different scales. In actual operation, changes in equipment temperature generally have characteristics such as periodicity and time scale differences. Through multi-scale domain feature extraction, it is possible to Provide more comprehensive temperature characteristics, including short-term changes, periodic changes, and duration changes of the equipment, which can more accurately describe the status changes of the equipment and detect potential faults. At the same time, this method can also effectively reduce the impact of noise and improve monitoring accuracy and stability, thereby achieving better intelligent inspection results.

Figure 3 is a block diagram of the temperature feature extraction module in the intelligent power inspection system according to an embodiment of the present application. As shown in Figure 3, the temperature feature extraction module 120 includes: a first scale temperature feature extraction unit 121, configured to input the device temperature input vector into the first convolution layer of the multi-scale neighborhood feature extraction module. To obtain the first scale temperature feature vector, wherein the first convolution layer has a first one-dimensional convolution kernel of a first length; the second scale temperature feature extraction unit 122 is used to input the device temperature input vector The second convolution layer of the multi-scale neighborhood feature extraction module is used to obtain the second scale temperature feature vector, wherein the second convolution layer has a second one-dimensional convolution kernel of a second length, and the first The length is different from the second length; and, a multi-scale cascade unit 123 is used to concatenate the first-scale temperature feature vector and the second-scale temperature feature vector to obtain the device temperature feature vector.

More specifically, in the power intelligent inspection system 100, the first scale temperature feature extraction unit 121 is used to: use the first convolution layer of the multi-scale neighborhood feature extraction module to use the following first convolution formula Perform convolution coding on the device temperature input vector to obtain the first scale temperature feature vector; wherein the first convolution formula is:

Among them, a is the width of the first convolution kernel in the x direction, F(a) is the first convolution kernel parameter vector, G(xa) is the local vector matrix that operates with the first convolution kernel function, and w is the first convolution kernel parameter vector. The size of a convolution kernel, X represents the device temperature input vector, and Cov ₁ (X) is the first scale temperature feature vector.

More specifically, in the power intelligent inspection system 100, the second scale temperature feature extraction unit 122 is further configured to use the second convolution layer of the multi-scale neighborhood feature extraction module to perform the following second convolution: The formula performs convolution coding on the device temperature input vector to obtain the second scale temperature feature vector; wherein, the second convolution formula is:

Among them, b is the width of the second convolution kernel in the x direction, F(b) is the second convolution kernel parameter vector, G(xb) is the local vector matrix that operates with the second convolution kernel function, and m is the second convolution kernel parameter vector. The size of the two convolution kernels, X represents the device temperature input vector, and Cov ₂ (X) is the second scale temperature feature vector.

In the power intelligent inspection system 100, the S transformation module 130 is used to perform S transformation on the equipment vibration signal to obtain an S transformation time-frequency diagram. S transform is a time-frequency analysis method that divides the signal into a series of short-time windows and performs Fourier transform on each window to obtain the spectrum information of the signal in each time period. The S transform can provide local information of the signal in time and frequency, so it has been widely used in the field of signal processing. The S-transform time-frequency diagram is the result of the S-transform, which combines the time domain and frequency domain information of the signal and can be used to analyze the changes in time and frequency of the signal. It should be understood that since the vibration signal is an unsteady signal with time as the independent variable, it is often difficult to obtain comprehensive information through direct time domain analysis. However, by converting the time domain signal into the time-frequency domain through S transformation, it can simultaneously reflect The signal changes in time and frequency. Thus, while retaining the information of the original signal, the time-frequency characteristics of the vibration signal become more intuitive and clear, which is more conducive to subsequent diagnosis and processing.

Specifically, in the power intelligent inspection system 100, the S transformation module 130 is used to: perform S transformation on the equipment vibration signal using the following transformation formula to obtain the S transformation time-frequency diagram; wherein, the transformation The formula is:

Among them, s(f,τ) represents the S-transform time-frequency diagram, τ is the time shift factor, x(t) represents the vibration signal of the equipment, f represents the frequency, and t represents the time.

In the intelligent power inspection system 100, the vibration encoding module 140 is used to pass the S-transform time-frequency diagram through a convolutional neural network model as a filter to obtain the equipment vibration feature vector. Those of ordinary skill in the art should know that the convolutional neural network is a deep learning model that can effectively extract image features. It can extract features and reduce the dimensionality of the input image through operations such as convolution layers and pooling layers, thereby obtaining More compact feature vectors. In the technical solution of this application, the S-transform time-frequency diagram can be regarded as a kind of non-image-like data. By inputting it into the convolutional neural network, the characteristic vector of the equipment vibration signal can be obtained, thereby realizing electric power Equipment status diagnosis and fault prediction.

In the intelligent power inspection system 100, the fusion module 150 is used to fuse the equipment temperature feature vector and the equipment vibration feature vector to obtain a classification feature matrix. It should be understood that in actual operation, the working status of equipment is usually affected by a variety of factors, such as temperature, vibration and other parameters, among which there is a certain correlation and interdependence between different parameters. Integrating the two key characteristic information of temperature and vibration can further improve the accuracy and precision of equipment fault diagnosis, thereby better completing intelligent inspection of power.

It should be understood that, considering that the dimensions and number of samples of the device temperature feature vector and the device vibration feature vector are different, there may be some repeated information or some information irrelevant to the target task in the process of fusing the two feature vectors. , which leads to the problem of noise and redundant information during the fusion process of the equipment temperature feature vector and the equipment vibration feature vector. This information may have an adverse impact on the training of the model and affect the generalization ability of the model.

Figure 4 is a block diagram of a fusion module in the intelligent power inspection system according to an embodiment of the present application. As shown in Figure 4, the fusion module 150 includes: a sparse feature vector generation unit 151 for sparsely encoding the device temperature feature vector and the device vibration feature vector to obtain a first sparse feature vector and a second sparse feature vector. Sparse feature vector; the first JS divergence calculation unit 152 is used to calculate the first JS divergence of the first sparse feature vector relative to the second sparse feature vector; the second JS divergence calculation unit 153 is used to calculate Calculate the second JS divergence of the second sparse feature vector relative to the first sparse feature vector; the normalization processing unit 154 is used to perform the first JS divergence and the second JS divergence. Normalization processing to obtain the normalized first JS divergence and the normalized second JS divergence; and, a classification feature matrix generating unit 155 for generating the normalized first JS divergence and the normalized The second JS divergence is normalized as a weight, and the first sparse feature vector and the second sparse feature vector are fused to obtain the classification feature matrix.

Specifically, in the power intelligent inspection system 100, the first JS divergence calculation unit 152 is used to calculate the first sparse feature vector relative to the second sparse feature using the following first JS divergence formula: The first JS divergence of the vector; wherein, the first JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₁ represents the first JS divergence.

Specifically, in the power intelligent inspection system 100, the second JS divergence calculation unit 153 is used to calculate the second sparse feature vector relative to the first sparse feature using the following second JS divergence formula: The second JS divergence of the vector; where the second JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₂ represents the second JS divergence.

In one embodiment of the present application, the device temperature feature vector and the device vibration feature vector are sparsely encoded using a dictionary learning-based technology to obtain a first sparse feature vector and a second sparse feature vector. It should be understood that the main idea of technology based on dictionary learning is to learn a sparse representation so that the original feature vector can be described with a small number of non-zero weights, and these weights are calculated based on a dictionary defined in advance. Specifically, the technology based on dictionary learning performs sparse encoding on the device temperature feature vector and the device vibration feature vector to obtain the first sparse feature vector and the second sparse feature vector, including: defining a sparse encoder that can receive A feature vector and outputs a sparse encoding of the received feature vector. This encoding satisfies the following conditions: the received feature vector consists of a small number of non-zero values, which correspond to the basis vectors in the dictionary defined in advance. Establish a dictionary learning module. The input of this module is a set of training feature vectors, and the output is a set of base vectors. These base vectors constitute the dictionary. The goal of dictionary learning is to minimize the reconstruction error, that is, use the basis vectors in the dictionary to reconstruct the original feature vector so that the reconstruction error is minimized. Given a training set, train using sparse encoder and dictionary learning module. During the training process, the original feature vector is first passed to the sparse encoder to obtain sparse encoding. The basis vectors in the sparse encoding and dictionary are then input into the decoder to minimize the reconstruction error. This process is repeated until the model converges. For the new feature vector to be encoded, the sparse encoding associated with it is calculated using the already trained sparse encoder and dictionary.

The above feature distribution fusion algorithm uses the idea of sparse coding to effectively capture the structure and pattern information between two feature distributions without being affected by noise and redundant information, thereby improving the feature fusion effect. In this way, it can effectively reduce The information loss in the feature fusion process retains the important information in the original feature distribution, improves the quality and credibility of the data after feature fusion, and at the same time, can also effectively reduce the data dimension after feature fusion, reduce data redundancy and noise, and Effectively enhance the data expression ability after feature fusion, extract more implicit information and potential patterns, and improve the ability and level of data mining and knowledge discovery. Thus, the accuracy of the classification result obtained by the classification feature matrix through the classifier is improved.

In the power intelligent inspection system 100, the operating status result generation module 160 is used to pass the classification feature matrix through a classifier to obtain a classification result. The classification result is used to indicate whether the operating status of the power equipment is normal. In this way, the operating status of power equipment can be judged more accurately, avoiding the risk of human misjudgment, while also improving the efficiency of power inspections.

Figure 5 is a block diagram of the operating status result generation module in the intelligent power inspection system according to an embodiment of the present application. As shown in Figure 5, the running status result generation module 160 includes: an expansion unit 161, used to expand the classification feature matrix into a classification feature vector based on row vectors or column vectors; a fully connected encoding unit 162, used to The classification feature vector is fully connected using the fully connected layer of the classifier to obtain the coded classification feature vector; and, the classification result generation unit 163 is used to pass the coded classification feature vector through the Softmax of the classifier. Classification function to obtain the classification results.

To sum up, the intelligent power inspection system 100 based on the embodiment of the present application is clarified, which extracts features from the temperature values and vibration signals of the power equipment during operation, and fuses the temperature features and vibration features to determine the Whether the electrical equipment is operating normally. In this way, the operating status of the power equipment can be judged more accurately, potential faults can be discovered early, and the normal operation of the power equipment can be ensured.

As mentioned above, the smart power inspection system 100 according to the embodiment of the present application can be implemented in various terminal devices, such as servers used for smart power inspection. In one example, the intelligent power inspection system 100 according to the embodiment of the present application can be integrated into a terminal device as a software module and/or a hardware module. For example, the smart power inspection system 100 may be a software module in the operating system of the terminal device, or may be an application program developed for the terminal device; of course, the smart power inspection system 100 may also be One of the many hardware modules of this terminal device.

Alternatively, in another example, the smart power inspection system 100 and the terminal device may also be separate devices, and the smart power inspection system 100 may be connected to the terminal device through a wired and/or wireless network, and Transmit interactive information according to the agreed data format.

Example methods

Figure 6 is a flow chart of an intelligent power inspection method according to an embodiment of the present application. As shown in Figure 6, the intelligent power inspection method includes: S110, obtaining equipment temperature values at multiple predetermined time points within a predetermined time period and equipment vibration signals within the predetermined time period collected by vibration sensors; S120 , arrange the equipment temperature values of the multiple predetermined time points into equipment temperature input vectors according to the time dimension and then pass the multi-scale neighborhood feature extraction module to obtain the equipment temperature feature vector; S130, perform S transformation on the equipment vibration signal to Obtain the S transform time-frequency diagram; S140, pass the S transform time-frequency diagram through the convolutional neural network model as a filter to obtain the equipment vibration feature vector; S150, combine the equipment temperature feature vector and the equipment vibration feature vector Fusion is performed to obtain a classification feature matrix; and, the classification feature matrix is passed through a classifier to obtain a classification result, where the classification result is used to indicate whether the operating status of the power equipment is normal.

In one example, in the above-mentioned intelligent power inspection method, the equipment temperature values at the plurality of predetermined time points are arranged into equipment temperature input vectors according to the time dimension and then passed through a multi-scale neighborhood feature extraction module to obtain the equipment temperature. Feature vector, including: inputting the temperature input vector into the first convolution layer of the multi-scale neighborhood feature extraction module to obtain a first scale temperature feature vector, wherein the first convolution layer has a first length a first one-dimensional convolution kernel; input the temperature input vector into the second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale temperature feature vector, wherein the second convolution layer has a A second one-dimensional convolution kernel of two lengths, the first length being different from the second length; and concatenating the first scale temperature feature vector and the second scale temperature feature vector to obtain the The device temperature characteristic vector.

In one example, in the above-mentioned intelligent power inspection method, the temperature input vector is input into the first convolution layer of the multi-scale neighborhood feature extraction module to obtain a first-scale temperature feature vector, wherein: The first convolution layer has a first one-dimensional convolution kernel of a first length, including: using the first convolution layer of the multi-scale neighborhood feature extraction module to input the device temperature with the following first convolution formula The vector is convolutionally encoded to obtain the first scale temperature feature vector; where the first convolution formula is:

Among them, a is the width of the first convolution kernel in the x direction, F(a) is the first convolution kernel parameter vector, G(xa) is the local vector matrix that operates with the first convolution kernel function, and w is the first convolution kernel parameter vector. The size of a convolution kernel, X represents the device temperature input vector, and Cov ₁ (X) is the first scale temperature feature vector.

In one example, in the above intelligent power inspection method, the temperature input vector is input into the second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale temperature feature vector, where: The second convolution layer has a second one-dimensional convolution kernel of a second length, the first length being different from the second length, including: using the second convolution layer of the multi-scale neighborhood feature extraction module The device temperature input vector is convolutionally encoded with the following second convolution formula to obtain the second scale temperature feature vector; wherein the second convolution formula is:

Among them, b is the width of the second convolution kernel in the x direction, F(b) is the second convolution kernel parameter vector, G(xb) is the local vector matrix that operates with the second convolution kernel function, and m is the second convolution kernel parameter vector. The size of the two convolution kernels, X represents the device temperature input vector, and Cov ₂ (X) is the second scale temperature feature vector.

In one example, in the above-mentioned intelligent power inspection method, performing S-transform on the equipment vibration signal to obtain an S-transform time-frequency diagram includes: performing S-transform on the equipment vibration signal using the following transformation formula to obtain The S-transform time-frequency diagram; wherein, the transformation formula is:

Among them, s(f,τ) represents the S-transform time-frequency diagram, τ is the time shift factor, x(t) represents the vibration signal of the equipment, f represents the frequency, and t represents the time.

In one example, in the above-mentioned intelligent power inspection method, fusing the equipment temperature feature vector and the equipment vibration feature vector to obtain a classification feature matrix includes: merging the equipment temperature feature vector and the equipment vibration feature vector. The device vibration feature vector is sparsely encoded to obtain a first sparse feature vector and a second sparse feature vector; the first JS divergence of the first sparse feature vector relative to the second sparse feature vector is calculated; the second sparse feature vector is calculated. The second JS divergence of the sparse feature vector relative to the first sparse feature vector; normalize the first JS divergence and the second JS divergence to obtain the normalized first JS divergence and the normalized second JS divergence; and, using the normalized first JS divergence and the normalized second JS divergence as weights, fuse the first sparse feature vector and the second Sparse feature vectors to obtain the classification feature matrix.

In one example, in the above intelligent power inspection method, calculating the first JS divergence of the first sparse feature vector relative to the second sparse feature vector includes: using the following first JS divergence formula Calculate the first JS divergence of the first sparse feature vector relative to the second sparse feature vector; wherein the first JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₁ represents the first JS divergence.

In one example, in the above intelligent power inspection method, calculating the second JS divergence of the second sparse feature vector relative to the first sparse feature vector includes: using the following second JS divergence formula Calculate the second JS divergence of the second sparse feature vector relative to the first sparse feature vector; wherein the second JS divergence formula is:

Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₂ represents the second JS divergence.

In one example, in the above intelligent power inspection method, passing the classification feature matrix through a classifier to obtain a classification result, the classification result being used to indicate whether the operating status of the power equipment is normal includes: passing the classification feature matrix The classification feature matrix is expanded into a classification feature vector based on a row vector or a column vector; the classification feature vector is fully connected using a fully connected layer of the classifier to obtain a coded classification feature vector; and, the coded classification feature is The vector is passed through the Softmax classification function of the classifier to obtain the classification result.

In summary, the intelligent power inspection method based on the embodiments of the present application has been clarified, which extracts features from the temperature values and vibration signals of the power equipment when it is running, and fuses the temperature features and vibration features to determine the power status of the power equipment. Whether the equipment is operating normally. In this way, the operating status of the power equipment can be judged more accurately, potential faults can be discovered early, and the normal operation of the power equipment can be ensured.

Example electronic device

Next, an electronic device according to an embodiment of the present application is described with reference to FIG. 7 .

Figure 7 is a block diagram of an electronic device according to an embodiment of the present application.

As shown in FIG. 7 , the electronic device 10 includes one or more processors 11 and memories 12 .

The processor 11 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

Memory 12 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and/or cache memory (cache). The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions can be stored on the computer-readable storage medium, and the processor 11 can run the program instructions to implement the intelligent power inspection method and/or the various embodiments of the present application described above. or other desired functionality. Various contents such as power equipment temperature values, power equipment vibration signals, etc. can also be stored in the computer-readable storage medium.

In one example, the electronic device 10 may further include an input device 13 and an output device 14, and these components are interconnected through a bus system and/or other forms of connection mechanisms (not shown).

The input device 13 may include, for example, a keyboard, a mouse, and the like.

The output device 14 can output various information to the outside, including the results of judging whether the power equipment is operating normally. The output device 14 may include, for example, a display, a speaker, a printer, a communication network and remote output devices connected thereto, and the like.

Of course, for simplicity, only some of the components in the electronic device 10 related to the present application are shown in FIG. 7 , and components such as buses, input/output interfaces, etc. are omitted. In addition, the electronic device 10 may also include any other appropriate components depending on the specific application.

Example computer program products and computer-readable storage media

In addition to the above-mentioned methods and devices, embodiments of the present application may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to execute the “exemplary method” described above in this specification. The steps in the intelligent power inspection method according to various embodiments of the present application are described in the section.

The computer program product can be used to write program codes for performing the operations of the embodiments of the present application in any combination of one or more programming languages, including object-oriented programming languages, such as Java, C++, etc. , also includes conventional procedural programming languages, such as the "C" language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server execute on.

In addition, embodiments of the present application may also be a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are run by a processor, the computer program instructions cause the processor to execute the above-mentioned "example method" part of this specification. The steps in the intelligent power inspection method according to various embodiments of the present application are described in .

The computer-readable storage medium may be any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: electrical connection with one or more conductors, portable disk, hard disk, random access memory (RAM), read only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.

Claims (10)

1. An intelligent power inspection system, characterized by including: An equipment data acquisition module, used to obtain equipment temperature values at multiple predetermined time points within a predetermined time period and equipment vibration signals within the predetermined time period collected by vibration sensors; A temperature feature extraction module, configured to arrange the device temperature values at the plurality of predetermined time points into device temperature input vectors according to the time dimension and then pass the multi-scale neighborhood feature extraction module to obtain the device temperature feature vector; S transformation module, used to perform S transformation on the equipment vibration signal to obtain S transformation time-frequency diagram; A vibration coding module, used to pass the S-transformed time-frequency diagram through a convolutional neural network model as a filter to obtain the equipment vibration feature vector; A fusion module, used to fuse the equipment temperature feature vector and the equipment vibration feature vector to obtain a classification feature matrix; and An operating status result generation module is used to pass the classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate whether the operating status of the power equipment is normal. 2. The intelligent power inspection system according to claim 1, characterized in that the temperature feature extraction module includes: The first scale temperature feature extraction unit is used to input the device temperature input vector into the first convolution layer of the multi-scale neighborhood feature extraction module to obtain the first scale temperature feature vector, wherein the first convolution The layer has a first one-dimensional convolution kernel of a first length; A second scale temperature feature extraction unit is configured to input the device temperature input vector into the second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale temperature feature vector, wherein the second convolution a layer having a second one-dimensional convolution kernel of a second length, the first length being different from the second length; and A multi-scale cascade unit is used to cascade the first-scale temperature feature vector and the second-scale temperature feature vector to obtain the device temperature feature vector. 3. The electric power intelligent inspection system according to claim 2, characterized in that the first scale temperature feature extraction unit is further configured to: use the first convolution layer of the multi-scale neighborhood feature extraction module to The following first convolution formula performs convolution coding on the device temperature input vector to obtain the first scale temperature feature vector; Wherein, the first convolution formula is: Among them, a is the width of the first convolution kernel in the x direction, F(a) is the first convolution kernel parameter vector, G(xa) is the local vector matrix that operates with the first convolution kernel function, and w is the first convolution kernel parameter vector. The size of a convolution kernel, X represents the device temperature input vector, and Cov ₁ (X) is the first scale temperature feature vector. 4. The intelligent electric power inspection system according to claim 3, characterized in that the second scale to-be-detected feature extraction unit is further configured to: use the second convolution layer of the multi-scale neighborhood feature extraction module. Perform convolution coding on the device temperature input vector with the following second convolution formula to obtain the second scale temperature feature vector; Wherein, the second convolution formula is: Among them, b is the width of the second convolution kernel in the x direction, F(b) is the second convolution kernel parameter vector, G(xb) is the local vector matrix that operates with the second convolution kernel function, and m is the second convolution kernel parameter vector. The size of the two convolution kernels, X represents the device temperature input vector, and Cov ₂ (X) is the second scale temperature feature vector. 5. The intelligent power inspection system according to claim 4, characterized in that the S transformation module is used to perform S transformation on the equipment vibration signal using the following transformation formula to obtain the S transformation time-frequency diagram. ; Among them, the transformation formula is: Among them, s(f,τ) represents the S-transform time-frequency diagram, τ is the time shift factor, x(t) represents the vibration signal of the equipment, f represents the frequency, and t represents the time. 6. The intelligent power inspection system according to claim 5, characterized in that the fusion module includes: A sparse feature vector generation unit configured to sparsely encode the device temperature feature vector and the device vibration feature vector to obtain a first sparse feature vector and a second sparse feature vector; A first JS divergence calculation unit, configured to calculate the first JS divergence of the first sparse feature vector relative to the second sparse feature vector; A second JS divergence calculation unit, configured to calculate the second JS divergence of the second sparse feature vector relative to the first sparse feature vector; A normalization processing unit, configured to normalize the first JS divergence and the second JS divergence to obtain a normalized first JS divergence and a normalized second JS divergence; and A classification feature matrix generating unit, configured to use the normalized first JS divergence and the normalized second JS divergence as weights to fuse the first sparse feature vector and the second sparse feature vector to Obtain the classification feature matrix. 7. The electric power intelligent inspection system according to claim 6, characterized in that the first JS divergence calculation unit is used to calculate the first sparse feature vector relative to the following first JS divergence formula. the first JS divergence of the second sparse feature vector; Among them, the first JS divergence formula is: Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₁ represents the first JS divergence. 8. The electric power intelligent inspection system according to claim 7, characterized in that the second JS divergence calculation unit is used to calculate the second sparse feature vector relative to the following second JS divergence formula. the second JS divergence of the first sparse feature vector; Among them, the second JS divergence formula is: Wherein, S ₁ is the first sparse feature vector, S ₂ is the second sparse feature vector, S is the average distribution of the first sparse feature vector and the second sparse feature vector, and KL represents the KL divergence. , JSD ₂ represents the second JS divergence. 9. The intelligent power inspection system according to claim 8, characterized in that the operating status result generation module includes: An expansion unit configured to expand the classification feature matrix into a classification feature vector based on row vectors or column vectors; a fully connected coding unit, configured to use a fully connected layer of the classifier to perform fully connected encoding on the classification feature vector to obtain a coded classification feature vector; and A classification result generating unit is used to pass the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result. 10. An intelligent power inspection method, characterized by including: Obtaining device temperature values at multiple predetermined time points within a predetermined time period and equipment vibration signals within the predetermined time period collected by the vibration sensor; Arrange the equipment temperature values at the multiple predetermined time points into equipment temperature input vectors according to the time dimension and then pass the multi-scale neighborhood feature extraction module to obtain the equipment temperature feature vector; Perform S-transform on the equipment vibration signal to obtain an S-transform time-frequency diagram; Pass the S-transformed time-frequency diagram through the convolutional neural network model as a filter to obtain the equipment vibration feature vector; Fusion of the equipment temperature feature vector and the equipment vibration feature vector to obtain a classification feature matrix; and The classification feature matrix is passed through a classifier to obtain a classification result, which is used to indicate whether the operating status of the power equipment is normal.