DE102021117283A1 - Method and computer program for determining the sensor quality of an optical sensor system - Google Patents

Abstract

The invention relates to a method for determining the sensor quality of an optical sensor system in relation to a fiber material of a fiber composite material comprising the fiber material and a matrix material embedding the fiber material, the optical sensor system for detecting a fiber material surface formed by the fiber material and for creating a depth image based on the detected fiber material surface is formed, with the steps:- providing a digital material model of the fiber material in a computing unit, wherein the material model describes an optical reflection behavior of the fiber material by at least one reflection-related parameter;- initializing a digital sensor model of the optical sensor system as a function of at least one sensor-related parameter of the optical sensor system and the reflection-related parameter of the fiber material modeled by the material model provided by the rake nunit;- Calculation of a probability distribution by means of a probability density function for at least one parameter of the digital sensor model describing the optical sensor system by the computing unit;- Calculation of an estimate of a best possible parameter value for at least one parameter for which a probability distribution was calculated, depending on the calculated probability distribution and an information inequality by the computing unit;- determining the sensor quality of the optical sensor system as a function of at least one calculated estimate of the best possible parameter value by the computing unit.

Description

The invention relates to a method for determining the sensor quality of an optical sensor system in relation to a fiber material of a fiber composite material comprising the fiber material and a matrix material embedding the fiber material, the optical sensor system for detecting a fiber material surface formed by the fiber material and for creating a depth image based on the detected Fiber material surface is formed. The invention also relates to a computer program for this purpose.

The invention also relates to a method for depositing fiber material of a fiber composite material comprising the fiber material and a matrix material embedding the fiber material on a tool surface, with an optical sensor system detecting a fiber material surface of the deposited fiber material, creating a depth image based on the detected fiber material surface and defects on the deposited Fiber material can be detected depending on the created depth image. The invention also relates to a plant for laying down fiber material for this purpose.

Due to the weight-specific strength and rigidity of fiber composite components that are made from a fiber composite material, such components in aerospace and many other areas of application, such as the automotive sector, are nowadays almost indispensable. In the production of a fiber composite component, a matrix material embedding the fiber material is usually cured under the application of temperature and pressure and thus forms an integral unit with the fiber material after curing. The reinforcing fibers of the fiber material are thus forced in their specified direction and can transfer the loads that occur in the specified direction.

Fiber composite materials from which such fiber composite components are produced generally have two main components, namely a fiber material on the one hand and a matrix material on the other. In addition, other secondary components can also be used, such as binder materials or additional functional elements that are to be integrated into the component. If dry fiber materials are provided for the production, the matrix material of the fiber composite material is infused into the fiber material by an infusion process during the production process, through which the dry fiber material is impregnated with the matrix material. This usually happens because of a pressure difference between the matrix material and the fiber material, for example by the fiber material being evacuated using a vacuum pump. In contrast to this, fiber composite materials are also known in which the fiber material is already pre-impregnated with the matrix material (so-called prepregs).

Before the matrix material hardens, the fiber material is usually introduced into a mold that reproduces the subsequent component shape with its shaping tool surface. Both dry and pre-impregnated fiber materials can be placed or introduced into the mold. For the production of large-scale structural components, such as wing shells of commercial aircraft or rotor blades of wind turbines, automated fiber laying processes are used to optimize the laying process, in which a production system and at least one fiber laying head are used to lay a quasi-endless fiber material fed to the fiber laying head on the tool. With the so-called fiber placement technology, for example, pre-impregnated fiber materials are placed in webs on the mold using such a fiber laying head. The fiber laying head is arranged on a robot and can be moved or moved relative to the mold. As a result, the individual fiber webs can first be laid down in webs and then in layers on the tool surface. With fiber placement technology, several, usually 8, 16 or 32 narrow material strips, so-called tows, are placed on the tool at the same time. In contrast, with fiber tape laying technology, wide fiber webs, also known as tapes (usually 150 mm, 300 mm or 600 mm wide with a thickness of a few tenths of a millimeter), are laid on the forming tool using the fiber laying head.

Such an automated fiber laying system is, for example, from DE 10 2010 015 027 B1 known in which several robots are guided on a revolving rail system, each of which has a laying head as an end effector. Fiber material is continuously fed to the laying heads from a fiber magazine by means of a fiber feed, while the individual robots place the fiber material supplied to them with their laying heads on a mold provided in the center of the circulating rail system. With the help of the fiber laying heads, so-called fiber laminates, also known as fiber preforms, are produced in the mold, in which the process of curing the matrix material has not yet taken place. Such fiber laminates (also called fiber preforms) thus form the structure part in a raw state before the matrix material hardens. In this way, dry fiber materials can be laid down, which are infused with a matrix material at a later point in time. For this purpose, a vacuum structure is created under which the dry fiber preform is enclosed, with the dry fiber preform then being evacuated. However, it is also conceivable that fiber materials that have already been pre-impregnated are laid down, so that an infusion of matrix material can be omitted.

Fiber laminates or fiber preforms for large-scale structural components, such as wings or rotor blades, often consist of many individual layers, which can include more than 100 layers depending on the area of application of the component. In order to ensure high component quality, the number of process errors that occur when laying down the fiber material must be minimized or prevented entirely. The occurrence of process errors or manufacturing deviations with a major negative impact on the strength of the component structure must, if possible, be fully recognized in the process and, if possible, also prevented.

In practice, manual inspection of the fiber laminates or the individual layers has often been established for this purpose. Depending on the size of the component, the availability of employees and the frequency of errors within an individual layer, it takes an average of 15 to 30 minutes to examine the placement of a single fiber layer. However, since the laying of a single fiber layer can usually be carried out much more quickly using automated fiber laying systems, the system efficiency is significantly reduced by manual inspection. For example, up to 400 individual layers are required to manufacture wing shells for short-haul aircraft, resulting in a plant downtime of 6,000 to 12,000 minutes per component due to the manual inspection.

Likewise, the inspection by a trained employee is highly dependent on the level of experience and, moreover, is not qualitatively sufficient to the extent that the manufacturing tolerances actually require. As an example, a permissible distance of 1 mm between 2 webs of material is given, which can only be qualitatively examined using tools and this can only be spot-checked.

For these reasons, an automated examination of laid fiber layers to detect laying defects is desirable in order to improve the detection of surface defects in laid fiber materials both qualitatively and quantitatively.

From the DE 10 2013 104 546 A1 a method for detecting defects in semi-finished fiber products placed on a tool surface is known, using a light projection method in which the surface is illuminated with light from a light source from a first direction and the light reflected from the surface is recorded with a camera from another, second direction a height profile of the fiber material surface is generated as a function of the image data recorded by the camera. Based on this height profile, flaws can now be detected by an image evaluation in the fiber material surface.

From the DE 10 2013 112 260 A1 a method for detecting defects using a light projection method is also known, in which case the fiber material surface is illuminated by means of a lighting unit in addition to the light from the light source of the light projection method and then, in addition to the height profile, the defects are also determined as a function of an intensity distribution of the reflected light of the lighting unit .

Furthermore, from the DE 10 2013 104 545 A1 a method for detecting defects using a light projection method is known, in which case, in addition to determining the height profile, the tool surface or the deposited fiber materials are tempered and then recorded with a thermal camera. Based on the thermographic images of the thermographic camera and the underlying height profile from the light projection process, defects on the fiber material surface can now be detected.

The quality assurance systems described above as examples, with which quality-related material properties of fiber materials of a fiber composite material can be determined during the manufacture of a fiber composite component, are fundamentally based on the idea of capturing the fiber material surface using a laser light section sensor and a height profile (depth image, usually shown in false colors) of the laid fiber material to create, from which quality-related material properties can then be determined.

With quality assurance systems of this type, with which the laying quality of laid fiber materials can be determined in particular within a fiber laying process during fiber laying (so-called inline quality assurance), there is a need to be able to assess the quality or accuracy of the quality assurance system in advance. For structurally critical areas, there is also a requirement for proof ses for the functionality of the quality assurance system, so that a manual inspection of the manufactured fiber laminates can be dispensed with. Proof of process-reliable detection is necessary in order to make automated depositing of fiber material and automated quality assurance of deposited fiber material certifiable, particularly in aviation.

From the post-published DE 10 2020 103 067 A1 a method for verifying an automated quality assurance system for detecting at least one quality-related material property of deposited fiber materials of a fiber composite material for the production of a fiber composite component is known, wherein the quality assurance system to be verified acquires measurement data of a fiber material surface of the deposited fiber material by means of at least one sensor of a first recording sensor system, from the measurement data a plurality of measurement data characteristics determined by means of a determination unit and recognizes the at least one quality-related material property from the measurement data features by means of an evaluation unit. Various series of measurements are carried out on a fiber material sample using different measuring devices, with the quality assurance system to be verified being able to be verified with regard to its measurement quality by means of a provided, machine-learned correlation of measurement data from the different measuring devices.

It is the object of the present invention to specify an improved method with which the sensor quality of an optical sensor system, in particular a light section sensor, can be determined simply and quickly during the quality assurance of fiber material of a fiber composite material.

The object is achieved with the method according to claim 1 according to the invention. Advantageous configurations of the invention can be found in the relevant documents.

According to claim 1, a method for determining the sensor quality of an optical sensor system in relation to a fiber material is proposed, the optical sensor system being designed to detect a fiber material surface formed by the fiber material and to create a depth image based on the detected fiber material surface. The optical sensor system is therefore a quality assurance system, as is known from the prior art and is used for automated quality control when laying down fiber material to produce a fiber composite component. Such an optical sensor system has at least one optical sensor (camera) that can record or detect the fiber material surface in a predetermined spectrum. In addition, it is conceivable that the optical sensor system has additional lighting elements in order to irradiate the fiber material surface.

Such a quality assurance system can be a laser light section sensor, for example, which projects a laser line onto the fiber material surface of the fiber material and records the projected laser line by means of a camera. Since both the laser line and the camera are at different angles to the fiber material surface, in particular the laser line is not orthogonal to the fiber material surface, unevenness from a deviation of the straight laser line can be automatically detected by image evaluation. A depth image can then be created from this, which, for example, encodes depth information in the two-dimensional representation of the fiber material surface by means of false-color representations. Such a depth image therefore represents a height profile of the fiber material surface of the laid fiber material.

The fiber material is a component of a fiber composite material that was placed on a mold to produce a fiber composite component. The fiber material can be dry or pre-impregnated (prepreg) fiber material. The fiber material can be a large number of tapes or tows.

First, a digital material model of the fiber material is provided in a computing unit, the material model describing an optical reflection behavior of the fiber material using at least one reflection-related parameter. Such a reflection-related parameter can be, for example, the reflected light intensity of a laser irradiation of the fiber material or the fiber material surface.

Based on this digital material model, a digital sensor model of the optical sensor system is now initialized as a function of at least one sensor-related parameter of the optical sensor system. Based on the knowledge of the input and processing, the digital sensor model abstracts the digital response of the sensor and thus reproduces its behavior in an abstract form. Such a sensor model can be mapped according to the EMVA 1288 sensor model standard, for example.

For example, based on a calculation of the number of received photons and Parameters of an internal sensor characteristic (camera characteristic), the sensor response is modeled, e.g. in the form of a digital gray value for each camera pixel.

With the aid of the computing unit, a probability distribution is now calculated using a probability density function for at least one parameter of the digital sensor model describing the optical sensor system, from which an estimate of the best possible parameter value for the at least one parameter for which the probability distribution was calculated, based on an information inequality, is calculated can. Depending on the best possible parameter values calculated in this way, the sensor quality of the optical sensor system can then be determined in relation to the fiber material used.

The present invention therefore offers the possibility of modeling and evaluating the best achievable data quality of an optical sensor system for quality assurance in the production of fiber composite components, so that even before the use of such an optical sensor system it can be checked whether the physically best achievable data quality is sufficient for the requirement of the automated manufacturing process. In addition, it can be estimated at an early stage how close the desired image quality is to the physically possible optimum and thus also how high the effort required for optimization and parameterization of such an optical sensor system is. An advantageous selection of an optical sensor system for the fiber material used can thus also be determined at an early stage and thus potentially high follow-up costs can be saved.

The inventors have recognized that by modeling the reflection behavior of a fiber material surface of a fiber material and by modeling the sensor used and its given characteristics using a probability density function and an information inequality, the best achievable data quality can be determined in advance from a physical point of view, so in advance to be able to evaluate the optical sensor system used with regard to the data quality to be achieved.

Accordingly, the determination of the sensor quality is understood to mean the determination of the data quality to be achieved for the optical sensor system used in relation to the fiber material used.

According to one embodiment, it is provided that the optical sensor system is a laser light section sensor in which a line laser of the optical sensor system projects a laser line onto the fiber material surface, this projected laser line is detected by an optical sensor of the optical sensor system and by a processing unit depending on the detected laser line a depth image of the fiber material surface is created.

The laser line is projected onto the fiber material from a first direction, while the fiber material is recorded together with the projected laser line from a second direction, which differs from the first direction. In this case, the laser line is preferably not projected onto the fiber material surface orthogonally to the latter. With the help of an image evaluation, the distortion of the view of a straight laser line on the projected fiber material surface can now be detected and a depth image can then be created from this, which encodes depth information in the two-dimensional plane of the fiber material surface. This creates a height profile of the fiber material surface. Based on such a height profile, flaws on the fiber material surface, which are known from the prior art, can then be detected.

According to one embodiment, it is provided that the digital sensor model of the optical sensor system is initialized and modeled by the computing unit as a function of at least one sensor-related optical input parameter of the optical sensor system, at least one sensor-related processing parameter of the optical sensor system and/or at least one sensor-related output parameter of the optical sensor system.

Optical input parameters of the sensor model can be the number of photons, for example. Sensor-related processing parameters, on the other hand, relate to a sensory characteristic of the sensor system used and are usually specified by the manufacturer of the optical sensor system. A sensor-related output parameter of the optical sensor system can be worth a digital gray scale.

According to one embodiment it is provided that a geometric dimension (A) of an optical sensor pixel of the optical sensor system, an irradiance (E) of the back reflection from the fiber material surface, an exposure time (t), a wavelength (λ) of a light projection projected onto the fiber material surface and/or or an average number of photons (µ _p ) incident on the optical sensor system as sensor-related Parameter used to initialize the digital sensor model.

mit A als geometrische Abmessung eines optischen Sensorpixels, E als Bestrahlungsstärke der Rückreflexion von der Fasermaterialoberfläche, t als Belichtungszeit, λ als Wellenlänge einer auf die Fasermaterialoberfläche projizierten Lichtprojektion, h als das Plank'sche Wirkungsquantum und c als die Lichtgeschwindigkeit.According to one embodiment, the average number of photons (µ _p ) incident on the optical sensor system is calculated as follows: $${\mathrm{\mu }}_p=\frac{AEt\lambda }{Hc}$$

where A is the geometric dimension of an optical sensor pixel, E is the irradiance of the back reflection from the fiber material surface, t is the exposure time, λ is the wavelength of a light projection projected onto the fiber material surface, h is the Plank constant and c is the speed of light.

According to one embodiment, it is provided that the Fisher information is used as the probability density and the Fisher information matrix is used as the probability density function. The photons hitting the camera sensor are used as an input for the Fisher information matrix as a probability density function.

mit θ als Parametervektor mit wenigstens einem sensorbezogenen Parameter, k als Index des k-ten Pixel aller zuvor betrachteten Pixel der betrachteten Pixelumgebung des optischen Sensorsystems, µ_y als Mittelwert des digitalen Ausgangssignals des optischen Sensorsystems, µ_e als die Anzahl von in das optischen Sensorsystem einfallender Photonen im Betrachtungszeitraum und E als Erwartungswert, wobei das zugrundeliegende Parametermodell wie folgt beschrieben ist: $${\mathrm{\mu }}_y=K\left({\mathrm{\mu }}_e+{\mathrm{\mu }}_d\right)$$

mit µ_e als die Anzahl von in das optischen Sensorsystem einfallender Photonen im Betrachtungszeitraum, µ_d als Mittelwert des schwarzen Rauschens und K als Verstärkungsfaktor des Signals.According to one embodiment, the Fisher information matrix is calculated as follows: $$I\left(\theta \right)={\displaystyle \sum _{k=1}^K{\left(\frac{\partial {\mathrm{\mu }}_y}{\partial \theta }\right)}^T}\left(\frac{\partial {\mathrm{\mu }}_y}{\partial \theta }\right)E\left[{\left(\frac{\partial \text{ln}\left(\mathrm{\rho }\left({\mathrm{\mu }}_e\right)\right)}{\partial {\mathrm{\mu }}_y}\right)}^2\right]$$

with θ as a parameter vector with at least one sensor-related parameter, k as the index of the kth pixel of all previously considered pixels of the considered pixel environment of the optical sensor system, µ _y as the mean value of the digital output signal of the optical sensor system, µ _e as the number of in the optical sensor system incident photons in the observation period and E as the expected value, with the underlying parameter model being described as follows: $${\mathrm{\mu }}_y=K\left({\mathrm{\mu }}_e+{\mathrm{\mu }}_{\mathrm{i.e}}\right)$$

with µ _e as the number of incident photons in the optical sensor system in the observation period, µ _d as the mean value of the black noise and K as the amplification factor of the signal.

According to one embodiment, it is provided that the Cramer-Rao lower bound is calculated for the estimation of the best possible parameter value for at least one parameter. The Cramer-Rao Lower Bound specifies the best possible precision for the underlying sensor model and the sensor-related parameters in the parameter vector based on the sensor model used.

mit θ als Parametervektor mit wenigstens einem sensorbezogenen Parameter.According to one embodiment, the Cramer-Rao lower bound is calculated as follows: $$COv\left(\widehat{\theta }\right)\ge I^{-1}\left(\theta \right)$$

with θ as a parameter vector with at least one sensor-related parameter.

According to one embodiment, it is provided that the digital material model is an intensity distribution of the radiation reflected by the fiber material surface as a reflection-related parameter.

According to one embodiment, it is provided that the intensity distribution is modeled using a 2D normal distribution function.

The object is also achieved according to the invention with the computer program according to claim 13, the computer program being set up with program code means for carrying out the method described above when the computer program is executed on a data processing system.

The object is also achieved according to the invention with the method for depositing fiber material according to claim 14, with an optical sensor system detecting a fiber material surface of the deposited fiber material, creating a depth image based on the detected fiber material surface and defects in the deposited fiber material depending on the created Depth image are detected. According to the invention, the sensor quality of the optical sensor system is determined in relation to the fiber material used according to the method described above.

The object is also achieved according to the invention with the system for depositing fiber material according to claim 15, wherein the system has an optical sensor system that is designed to detect a fiber material surface formed by the fiber material and to create a depth image based on the detected fiber material surface. The system has a data processing system for determining the sensor quality of the optical sensor system, which is set up to carry out the method described above.

Such a system can include at least one movement machine (e.g. an Indust ry robots or articulated-arm robots) on which a placement head is arranged as an end effector. A fiber material is continuously fed into the laying head, which is then laid down on a tool. At least one optical sensor (camera) of the optical sensor system can be arranged on the laying head, the laying head being moved relative to the tool while the fiber material is being laid down and the fiber material already laid down being detected by the optical sensor. Such an optical sensor of the optical sensor system can be a laser light section sensor, for example.

• 1 schematische Darstellung eines Ablegekopfes mit einem Laserlichtschnittsensor im Nachlauf;
• 2 schematisch stark vereinfachte Darstellung des Verfahrensablaufes;
• 3 schematische Darstellung eines Sensormodells;
• 4 schematische Darstellung des Messaufbaus zur Ermittlung reflexionsbezogener Parameter des Fasermaterials;

The invention is explained in more detail by way of example on the basis of the attached figures. Show it:

• 1 schematic representation of a laying head with a laser light section sensor in the wake;
• 2 schematic, highly simplified representation of the process sequence;
• 3 schematic representation of a sensor model;
• 4 schematic representation of the measurement setup for determining reflection-related parameters of the fiber material;

1 shows the method known from the prior art for recording a height profile using a laser light section sensor 5 following a laying head 1, which has a pressure roller 2 at the lower end, with which a flat semi-finished fiber product 3 is to be laid on a shaping tool surface 4. The semi-finished fiber products to be laid can be dry rovings or tows, but also pre-impregnated prepregs.

A laser light section sensor 5 is arranged on the laying head 1 downstream of the laying head 1 . With the aid of a laser light source 6, a laser light line 7 is projected onto the part of the semi-finished fiber product 3 that has already been laid down. The projection of the laser light in the form of a laser light line 7 onto the semi-finished fiber product 3 takes place from a first direction.

The laser light section sensor 5 also has a camera 8 which is arranged at a defined distance from the laser light source 6 and which records the laser light line 7 projected onto the semi-finished fiber product 3 from a different, second direction at a predetermined and defined angle.

The camera 8 of the light section sensor 5 is connected to an image evaluation unit 9 which now determines a corresponding height profile as a function of the recorded image data and the light reflected from the laser light line 7 . Since the placement head 1 is moved continuously on the tool surface 4 together with the light section sensor 5, a large number of image data results. The image data and/or the height profiles determined from the image data per point in time can then be stored in a database 9a so that a complete three-dimensional height profile or surface structure of the surface of the semi-finished fiber product 3 results.

In addition, the corresponding characteristic flaws can be derived from the height profiles with the aid of the image evaluation unit 9, so that while the semi-finished fiber products 3 are being laid down on the tool surface 4, the quality assurance of the laying process follows in real time.

2 shows the process sequence according to the invention in a highly simplified schematic representation, as it is carried out on a data processing system 100. First, a digital material model 110 is provided, from which a digital sensor model 120 is then initialized. The so-called Fisher information is then calculated from this chain as a probability distribution 130, which is known from the statistical parameter estimation and describes the previously modeled sensor parameters via a distribution density function. From this, the Cramer-Rao Lower Bound 140 is then calculated, which indicates an estimate of the physically best possible parameter value for each of the parameters previously described in the sensor and material model or a selection thereof. From this, the sensor quality of the optical sensor system can now be calculated in step 150, as in 1 described, determine.

3 shows a sensor model of an optical CCD/CMOS camera sensor from the EMVA 1288 standard, which is used to show the calculation of the sensor quality in detail below. A suitable mathematical model is used here, which retains the relevant information but abstracts unnecessary or overly detailed information. Such an optical camera sensor can be part of a laser light section sensor, for example, for which a laser line is projected onto the fiber material.

The sensor is modeled as a black box, so that only knowledge about the inputs and outputs of the sensor and their processing including the influencing interference is required.

The output O of such a sensor model is the sum of the input I and the change change in the input signal (eg due to noise, amplification, quantization, etc.) N, with the variables I, N, O each being characterized by a distribution function with an expected value and a variance. It follows: O = I + N

The influences of the sensor on the input signal, which are described by N, are sensor characteristics that must be determined separately or are specified by a sensor manufacturer. These are therefore assumed to be given in the following and will not be considered in detail.

The average number of incident photons μ _p is shown in Equation 1 below.
Eq. 1
$${\mathrm{\mu }}_p=\frac{AEt\lambda }{Hc}$$

A is the geometric dimensions of a pixel on the LLSS camera sensor and E is the irradiance of the reflected laser light, which depends on the nature of the fiber composite material under consideration. Variable t describes the exposure/irradiation time of the observed pixel. The parameter λ is the wavelength of the laser light. The constants h are Plank's constant and c the speed of light.

It is then necessary to model the reflection behavior of the laser line of the laser light section sensor on the fiber composite material. It was recognized here that the complex modeling of a quasi-infinite laser line can be broken down into the mathematical description of a single laser point, since the laser line can be understood as a collection and concatenation of individual laser points. A corresponding measurement setup is given in 4 sketched. 4 shows a measuring device 10 in order to be able to determine a reflection-related parameter of the fiber material for the material model. For this purpose, the fiber material 3 placed on the tool 4, which is also intended to be detected with the laser light section sensor 5, is detected using a point laser sensor 11 and the corresponding data is evaluated.

The point laser sensor 11 has a point laser 12 which projects a laser point 13 onto the surface of the fiber material 3 . The projection direction of the laser point 13 is in particular orthogonal to the surface of the fiber material 3 in order to avoid distortions due to the projection directions.

This laser point 13 projected onto the surface of the fiber material 3 is now recorded using a camera 14 of the point laser sensor 11 and corresponding digital image data is generated. This digital image data of the recorded laser point 13 is then transmitted to an evaluation unit 15, which extracts the features of the projected laser point 13, initializes the corresponding sensor model and determines the sensor quality from the initialized sensor model.

This simple structure offers the possibility of describing the signal quality resulting from the reflected radiation. The method implicitly takes into account the scattering and reflection properties of the fiber material and also implicitly the beam propagation in the fiber composite material.

$$q=\frac{1}{2\pi {\sigma }_g}\text{exp}\left(-\frac{x^2+y^2}{2{\sigma }_g{}^2}\right)$$

Specifically, the profile of the reflected photons or the intensity distribution of the reflected radiation for a laser point is modeled using a corresponding 2D normal distribution function. For this purpose, the photon distribution function f is shown in Equation 2, and the point response function q in Equation 3. $$ƒ=q\left(x,y\right)$$

$$q=\frac{1}{2\pi {\sigma }_G}\text{ex}\left(-\frac{x^2+y^2}{2{\sigma }_G{}^2}\right)$$

The coordinates (x,y) describe the position of the observed object in the image plane of the sensor. σ _g describes the standard distribution of the imaged laser point.

Taking into account the knowledge about the appearance of the laser line or the way of modeling the sensor, the probability distribution can be described as a probability density function ρ through the reflected laser point (amount of photons striking the surface of the sensor) as in Equation 4. It was recognized here that the laser point is not only ideally reflected and imaged onto an image pixel of the camera sensor due to the diffuse reflection behavior of the fiber composite material, but rather onto an environment of several individual pixels. $$\mathrm{\rho }\left({\mathrm{\mu }}_e\right)=\frac{e^{-{\mathrm{\mu }}_y}}{\sqrt{\pi {\sigma }_k}}{\displaystyle \sum _{l=0}^{\infty }\frac{{\mathrm{\mu }}_y{}^l}{l\text{!}}}{e}^{\frac{-{\left({\mathrm{\mu }}_e-l-n_k\right)}^2}{2{\sigma }_k{}^2}}$$

The parameters η _k and σ _k describe the mean value and the standard deviation of the intensity distribution on the kth sensor pixel without Illumination of the kth pixel. The variable µ _y describes the underlying parameter model, which is shown in Equation 5.

The value µ _e indicates how many photons hit the sensor pixel during the observation period. This parameter µ _e is therefore in turn dependent on the interference of the LLSS and the parameter µ _p from Equation 1. The interference of the sensor mentioned can be simplified, based on the EMVA 1288 standard, for the LLSS, taking into account a single and constant wavelength of the Lasers can be summarized in a single efficiency parameter QE. The following therefore applies: µ _e = µ _p *QE. $${\mathrm{\mu }}_y=K\left({\mathrm{\mu }}_e+{\mathrm{\mu }}_{\mathrm{i.e}}\right)$$

The parameter µ _y is the mean value of the digital output signal, which summarizes the individual parameters of the sensor model. µ _d represents the average of the "black noise", i.e. the average number of photons that are registered by the sensor when no (light) radiation is incident on it. The K value is the overall system gain.

Using the information from Equation 4, the Fisher Information Matrix (FIM) proposed for the method according to the invention can subsequently be calculated. This is shown in Equation 6. $$I\left(\theta \right)={\displaystyle \sum _{k=1}^K{\left(\frac{\partial {\mathrm{\mu }}_y}{\partial \theta }\right)}^{\text{T}}\left(\frac{\partial {\mathrm{\mu }}_y}{\partial \theta }\right)E\left[{\left(\frac{\partial \text{ln}\left(\mathrm{\rho }\left({\mathrm{\mu }}_e\right)\right)}{\partial {\mathrm{\mu }}_y}\right)}^2\right]}$$

The index k describes the kth pixel of all previously considered K pixels of the pixel environment considered. In the contexts described above, it is assumed in each case that the reflection of the point laser from the fiber composite material forms a normal distribution on the image plane around a specific point of impact. Despite the diffuse scattering behavior of the fiber composite material, this assumption is tolerable for a laser point reflection, since the diameter of the point of impact of the laser is comparatively small and the strong diffuse scattering of the fiber composite material impinges outside of the sensor or the defined observation area anyway. Thus, this strong diffuse scattering only leads to a reduction in the amount of incident photons. Assuming a uniform distribution over the image plane of the sensor around a point of impact, this leads to a so-called Poisson distribution for the discretized, pixel-by-pixel observation on an imaging sensor chip such as the LLSS camera sensor.

In order to then determine the maximum possible quality of the input signal, the so-called Cramer-Rao-Lower-Bound (CRLB) is calculated. This gives a lower bound for the variance of the estimate of θ. This is described in equation 7. $$COv\left(\widehat{\theta }\right)\ge I^{-1}\left(\theta \right)$$

The parameter vector θ contains all relevant parameters of the inspection system that are to be estimated. In the concrete example case described here, θ = (A, K, t, E, QE, λ) ^T . For the individual parameters, the lower cabinet results from the diagonal elements of the inverse FIM. Equation 4 describes the distribution of these parameters. According to the CRLB relationship from Equation 7, θ̂ is to be understood as an independent estimated vector, whereby the reciprocal value of the FIM I of the parameter vector θ according to the CRLB inequality can at most become as large as the variance of the vector θ̂. The Cramer-Rao lower bound is calculated for a parameter θ _i . A large number of parameters θ _i then form the parameter vector with at least one sensor-related parameter.

This therefore describes the maximum possible data/information quality according to the underlying mathematical model of the LLSS and the reflected laser light. It is thus possible to approximate or determine the maximum performance of an inspection system using the mathematically best possible input information. It can thus be shown at an early stage whether a corresponding optical sensor is suitable for the inspection of a specific, given fiber material. If a sensor is suitable, it can be determined on the basis of the additional power or the available "buffer" how well or how complex the adjustable sensor parameters have to be optimized in order to achieve the desired signal quality.

Reference List

1

laying head

2

pinch roller

3

fiber semi-finished product

4

tool surface

5

Laser light section sensor

6

laser light source

7

laser light line

8th

camera

9

image evaluation unit

9a

Database

10

measuring device

11

point laser sensor

12

point laser

13

laser spot

14

camera

15

evaluation unit

100

data processing system

110

material model

120

digital sensor model

130

probability distribution

140

Estimation of the best possible parameter value

150

sensor quality

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• DE 102010015027 B1 [0006]
• DE 102013104546 A1 [0011]
• DE 102013112260 A1 [0012]
• DE 102013104545 A1 [0013]
• DE 102020103067 A1 [0016]

Claims (15)

Method for determining the sensor quality (150) of an optical sensor system in relation to a fiber material of a fiber composite material comprising the fiber material and a matrix material embedding the fiber material, the optical sensor system for detecting a fiber material surface formed by the fiber material and for creating a depth image based on the detected fiber material surface is trained, with the steps: - Providing a digital material model (110) of the fiber material in a computing unit, wherein the material model (110) describes an optical reflection behavior of the fiber material by at least one reflection-related parameter; - Initializing a digital sensor model of the optical sensor system as a function of at least one sensor-related parameter of the optical sensor system and the material model provided by the material model (110) - modeled reflection-related parameters of the fiber material by the computing unit; - calculation of a probability distribution (130) by means of a probability density function for at least one parameter of the optical sensor system describing digital sensor model by the computing unit; - calculating an estimate of a best possible parameter value (140) for at least one parameter for which a probability distribution (130) was calculated, depending on the calculated probability distribution (130) and an information inequality by the computing unit; - Determining the sensor quality (150) of the optical sensor system as a function of at least one calculated estimate of the best possible parameter value (140) by the computing unit. procedure after claim 1 , characterized in that the optical sensor system is a laser light section sensor (5), in which a line laser of the optical sensor system projects a laser line onto the fiber material surface, this projected laser line is detected by means of an optical sensor of the optical sensor system and by means of a processing unit depending on the detected Laser line creates a depth image of the fiber material surface. procedure after claim 1 or 2 , characterized in that the digital sensor model of the optical sensor system is initialized by the computing unit as a function of at least one sensor-related optical input parameter of the optical sensor system, at least one sensor-related processing parameter of the optical sensor system and/or at least one sensor-related output parameter of the optical sensor system. Method according to one of the preceding claims, characterized in that a geometric dimension (A) of an optical sensor pixel of the optical sensor system, an irradiance (E) of the back reflection from the fiber material surface, an exposure time (t), a wavelength (λ) on the fiber material surface projected light projection and/or an average number of photons (μ _p ) incident in the optical sensor system is used as a sensor-related parameter for the initialization of the digital sensor model. procedure after claim 4 , characterized in that the average number of photons (µ _p ) incident on the optical sensor system is calculated as follows: $${\mathrm{\mu }}_p=\frac{AEt\lambda }{Hc}$$

where A is the geometric dimension of an optical sensor pixel, E is the irradiance of the back reflection from the fiber material surface, t is the exposure time, λ is the wavelength of a light projection projected onto the fiber material surface, h is the Plank constant and c is the speed of light. Method according to one of the preceding claims, characterized in that the EMVA 1288 sensor model is used as the digital sensor model (120). Method according to one of the preceding claims, characterized in that the Fisher information is used as the probability density function and the Fisher information matrix is used as the probability density function. procedure after claim 7 , characterized in that the Fisher information matrix is calculated as follows: $$I\left(\theta \right)={\displaystyle \sum _{k=1}^K{\left(\frac{\partial {\mathrm{\mu }}_y}{\partial \theta }\right)}^T}\left(\frac{\partial {\mathrm{\mu }}_y}{\partial \theta }\right)E\left[{\left(\frac{\partial \text{ln}\left(\mathrm{\rho }\left({\mathrm{\mu }}_e\right)\right)}{\partial {\mathrm{\mu }}_y}\right)}^2\right]$$

with 8 as a parameter vector with at least one sensor-related parameter, k as the index of the k-th pixel of all previously considered K pixels of the considered pixel environment of the optical sensor system, µ _y as the mean value of the digital output signal of the optical sensor system, µ _e as the number of in the optical Sensor system of incident photons in the observation period and E as expected value, where the underlying parameter model is described as follows: $${\mathrm{\mu }}_y=K\left({\mathrm{\mu }}_e+{\mathrm{\mu }}_{\mathrm{i.e}}\right)$$

with µ _e as the number of incident photons in the optical sensor system in the observation period, µ _d as the mean value of the black noise and K as the amplification factor of the signal. Method according to one of the preceding claims, characterized in that the Cramer-Rao lower bound is calculated for the estimation of the best possible parameter value (140) for at least one parameter. procedure after claim 9 , characterized in that the Cramer-Rao lower bound for a parameter is calculated as follows: $$COv\left(\widehat{\theta }\right)\ge I^{-1}\left(\theta \right)$$

with 8 as a parameter vector with at least one sensor-related parameter. Method according to one of the preceding claims, characterized in that the digital material model (110) is an intensity distribution of the radiation reflected by the fiber material surface as a reflection-related parameter. procedure after claim 11 , characterized in that the intensity distribution is modeled using a 2D normal distribution function. Computer program with program code means set up to carry out the method according to one of Claims 1 until 12 when the computer program is executed on a data processing system (100). Method for depositing fiber material of a fiber composite material, comprising the fiber material and a matrix material embedding the fiber material on a tool surface (4), wherein a fiber material surface of the deposited fiber material is detected by means of an optical sensor system, a depth image is created based on the detected fiber material surface and defects in the deposited fiber material are detected Depending on the created depth image are detected, characterized in that the sensor quality (150) of the optical sensor system in relation to the fiber material used according to the method according to one of Claims 1 until 12 is determined. Plant for depositing fiber material of a fiber composite material comprising the fiber material and a matrix material embedding the fiber material on a tool surface (4), the plant having an optical sensor system which is used to detect a fiber material surface formed by the fiber material and to create a depth image based on the detected fiber material surface is formed, characterized in that the system has a data processing system (100) for determining the sensor quality (150) of the optical sensor system, which is used to carry out the method according to one of Claims 1 until 12 is set up.

Images