Intelligent processing and adjustment method and system for automobile parts

Abstract

The present application relates to the field of intelligent adjustment, and specifically discloses an intelligent processing adjustment method and system for automobile parts, which uses an artificial intelligence algorithm based on deep learning to obtain a detection image of a brake pad after grinding, and uses a convolutional neural network model to extract the implicit relationship of the grinding area and the implicit relationship of the brake pad as a whole, and further integrates the characteristic information of the two to comprehensively judge whether the grinder should stop grinding. In this way, the intelligent processing adjustment of automobile parts can be accurately performed to improve the production efficiency and quality level of automobile parts.

Description

Technical Field

[0001] The present application relates to the field of intelligent adjustment, and more specifically, to an intelligent processing adjustment method and system for automobile parts. Background Art

[0002] Brake pads are one of the auto parts. In the car's braking system, brake pads are the most critical safety parts. The quality of all braking effects is determined by brake pads. After forming, brake pads will have burrs, scratches, grooves and other unevenness. Most of the existing grinder adjustment controls are manually judged, which are not real-time and intelligent.

[0003] Therefore, an optimized intelligent processing and adjustment scheme for automotive parts is expected to obtain the inspection image of the brake pad after grinding and extract the implicit relationship of the grinding area and the implicit relationship of the brake pad as a whole, and further integrate them to obtain the classification result, which is used to indicate whether the grinder stops grinding. Summary of the invention

[0004] In order to solve the above technical problems, this application is proposed. The embodiment of the present application provides an intelligent processing and adjustment method for automobile parts and a system thereof, which uses an artificial intelligence algorithm based on deep learning to obtain the detection image of the brake pad after grinding, and uses a convolutional neural network model to extract the implicit relationship of the grinding area and the implicit relationship of the brake pad as a whole, and further integrates the feature information of the two to comprehensively judge whether the grinder stops grinding. In this way, the intelligent processing and adjustment of automobile parts can be accurately performed to improve the production efficiency and quality level of automobile parts.

[0005] According to one aspect of the present application, there is provided a method for intelligent processing and adjustment of automobile parts, which comprises:

[0006] Acquire a detection image of the polished brake pad;

[0007] The detection image of the polished brake pad is passed through a polishing area target detection network to obtain a polishing area of ​​interest image;

[0008] Passing the polished region of interest image through a first convolutional neural network model as a filter to obtain a polished feature vector;

[0009] Passing the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector;

[0010] Performing spatial simultaneous projection between sub-dimensions of a feature set on the polishing feature vector and the detection feature vector to obtain the classification feature vector; and

[0011] The classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinding machine stops grinding.

[0012] In the above-mentioned intelligent processing and adjustment method for automobile parts, the polishing area target detection network is CenterNet, ExtremeNet, RepPoints, YOLOv1 or FCOS.

[0013] In the above-mentioned intelligent processing and adjustment method for automobile parts, the polishing area of ​​interest image is passed through the first convolutional neural network model as a filter to obtain a polishing feature vector, including: using each layer of the first convolutional neural network model as a filter to perform the following on the input data in the forward transfer of the layer: convolution processing on the input data to obtain a convolution feature map; global mean pooling of each feature matrix of the channel dimension of the convolution feature map to obtain a pooled feature map; and nonlinear activation of the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the first convolutional neural network as a filter is the polishing feature vector, and the input of the first layer of the first convolutional neural network as a filter is the polishing area of ​​interest image.

[0014] In the above-mentioned intelligent processing and adjustment method for automobile parts, the detection image is passed through the second convolutional neural network model as a filter to obtain a detection feature vector, including: using each layer of the second convolutional neural network model as a filter to perform the following on the input data in the forward transfer of the layer: performing convolution processing on the input data to obtain a convolution feature map; performing global mean pooling on each feature matrix of the channel dimension of the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the second convolutional neural network as a filter is the detection feature vector, and the input of the first layer of the second convolutional neural network as a filter is the detection image.

[0015] In the above-mentioned intelligent processing and adjustment method for automobile parts, the polishing feature vector and the detection feature vector are subjected to spatial joint projection between the sub-dimensions of the feature set to obtain the classification feature vector, including: calculating the covariance matrix between the polishing feature vector and the detection feature vector; performing eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues ​​and multiple eigenvectors corresponding to the multiple eigenvalues; and arranging the multiple eigenvectors into the classification feature vector.

[0016] In the above-mentioned intelligent processing and adjustment method for automobile parts, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinder stops grinding, including: using multiple fully connected layers of the classifier to fully connect encode the corrected classification feature vector to obtain an encoded classification feature vector; and, passing the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0017] According to another aspect of the present application, there is provided an intelligent processing and adjustment system for automobile parts, comprising:

[0018] An image acquisition module, used to acquire a detection image of the polished brake pad;

[0019] A target detection module, used for passing the detection image of the polished brake pad through a polishing area target detection network to obtain a polishing area of ​​interest image;

[0020] A polishing feature extraction module, used for passing the polishing region of interest image through a first convolutional neural network model as a filter to obtain a polishing feature vector;

[0021] A detection feature extraction module, used for passing the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector;

[0022] A projection module, used for performing spatial joint projection between sub-dimensions of a feature set on the polishing feature vector and the detection feature vector to obtain the classification feature vector;

[0023] The classification result generating module is used to pass the classification feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinding machine stops grinding.

[0024] In the above-mentioned intelligent processing and adjustment system for automobile parts, the target detection module includes: the polishing area target detection network is CenterNet, ExtremeNet, RepPoints, YOLOv1 or FCOS.

[0025] In the above-mentioned intelligent processing and adjustment system for automobile parts, the polishing feature extraction module includes: using each layer of the first convolutional neural network model as a filter to perform the following on the input data in the forward transfer of the layer: convolution processing on the input data to obtain a convolutional feature map; global mean pooling on each feature matrix of the channel dimension of the convolutional feature map to obtain a pooled feature map; and nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the first convolutional neural network as a filter is the polishing feature vector, and the input of the first layer of the first convolutional neural network as a filter is the polishing area of ​​interest image.

[0026] In the above-mentioned intelligent processing and adjustment system for automobile parts, the detection feature extraction module includes: using each layer of the second convolutional neural network model as a filter to perform the following on the input data in the forward transfer of the layer: performing convolution processing on the input data to obtain a convolution feature map; performing global mean pooling on each feature matrix of the channel dimension of the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the second convolutional neural network as a filter is the detection feature vector, and the input of the first layer of the second convolutional neural network as a filter is the detection image.

[0027] In the above-mentioned intelligent processing and adjustment system for automobile parts, the projection module includes: a covariance unit, used to calculate the covariance matrix between the polishing eigenvector and the detection eigenvector; a decomposition unit, used to perform eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues ​​and multiple eigenvectors corresponding to the multiple eigenvalues; and an arrangement unit, used to arrange the multiple eigenvectors into the classification eigenvector.

[0028] In the above-mentioned intelligent processing and adjustment system for automobile parts, the classification result generation module includes: using multiple fully connected layers of the classifier to fully connect the corrected classification feature vector to obtain an encoded classification feature vector; and passing the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0029] According to another aspect of the present application, an electronic device is provided, comprising: a processor; and a memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the above-mentioned intelligent processing and adjustment method for automobile parts.

[0030] According to another aspect of the present application, a computer-readable medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the above-mentioned intelligent processing and adjustment method for automobile parts.

[0031] Compared with the prior art, the present application provides an intelligent processing and adjustment method and system for automobile parts, which uses an artificial intelligence algorithm based on deep learning to obtain the detection image of the brake pad after grinding, and uses a convolutional neural network model to extract the implicit relationship of the grinding area and the implicit relationship of the brake pad as a whole, and further integrates the characteristic information of the two to comprehensively judge whether the grinder should stop grinding. In this way, the intelligent processing and adjustment of automobile parts can be accurately performed to improve the production efficiency and quality level of automobile parts. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0033] Figure 1 This is an application scenario diagram of the intelligent processing and adjustment method for automobile parts according to an embodiment of the present application.

[0034] Figure 2 The present invention is a flowchart of an intelligent processing and adjustment method for automobile parts according to an embodiment of the present application.

[0035] Figure 3 Schematic diagram of the architecture of the intelligent processing and adjustment method for automobile parts according to an embodiment of the present application.

[0036] Figure 4 The present invention is a flowchart of obtaining a polishing feature vector by passing the polishing region of interest image through a first convolutional neural network model as a filter in the intelligent processing and adjustment method for automobile parts according to an embodiment of the present application.

[0037] Figure 5 It is a block diagram of an intelligent processing and adjustment system for automobile parts according to an embodiment of the present application.

[0038] Figure 6 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0039] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0040] Application Overview

[0041] As mentioned in the background technology, in the intelligent processing and adjustment of automobile parts, the brake pads will have burrs, scratches, grooves and other unevenness after production and molding. At present, most of the existing grinders are controlled manually, which sometimes causes injuries to personnel due to improper operation, and are not real-time and intelligent. Therefore, an optimized intelligent processing and adjustment solution for automobile parts is expected.

[0042] In response to the above technical problems, the applicant of the present application obtains the inspection image of the brake pad after grinding and extracts the implicit relationship of the grinding area and the implicit relationship of the brake pad as a whole, and further integrates them to obtain the classification result, which is used to indicate whether the grinder stops grinding.

[0043] Accordingly, in the technical solution of the present application, it is considered that by acquiring the detection image of the polished brake pad, the area of ​​the brake pad can be extracted quickly and accurately. By detecting the image, various information such as the shape, surface defects, color, etc. of the brake pad can be obtained, and this information can be passed to the subsequent processing module, so as to facilitate the comprehensive and in-depth analysis of the brake pad. Through the visual display of the image, the problems of the brake pad can be clearly observed and diagnosed, which helps to improve the efficiency and level of production management and quality control. With the help of advanced image processing technology, the automated processing and analysis of a large number of brake pad images can be realized, avoiding the errors and instability of manual operation and improving production efficiency and consistency.

[0044] 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 demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields.

[0045] Specifically, in the technical solution of the present application, first, a detection image of the brake pad after grinding is obtained. Next, considering that the grinding area in the image can be quickly and accurately detected by the target detection network, the complicated process of manually drawing the region of interest can be avoided, thereby improving processing efficiency and accuracy. For the case where the image size is large and the number is large, directly performing feature extraction on the entire image will lead to a sharp increase in the amount of calculation, reducing the response speed and stability of the system. By obtaining the region of interest, the scope and amount of feature extraction can be reduced, and the real-time performance and reliability of the system can be improved. There may be different areas and features in the brake pad after grinding. In order to improve the expressive power and discrimination of the features, it is necessary to perform feature extraction on the specific grinding area. Therefore, the detection image of the brake pad after grinding is passed through the grinding area target detection network to obtain the grinding area of ​​interest image, so that the features of the grinding area can be more accurately located and extracted.

[0046] Then, considering that there may be a variety of features and information in the polishing area, it is necessary to extract the most representative and discriminative features to describe the polishing status. Convolutional neural networks have the ability to automatically learn feature representation and model training. By taking the image of the polishing area of ​​interest as input, a more expressive and discriminative polishing feature vector can be obtained. By obtaining the polishing feature vector, complex and abstract image information can be converted into a simple and clear numerical vector, which can better reflect the state and characteristics of the polishing area, and help to achieve early warning and management of the grinding machine stopping polishing. Therefore, the image of the polishing area of ​​interest is passed through the first convolutional neural network model as a filter to obtain the polishing feature vector.

[0047] Next, considering that although the image of the polishing area of ​​interest can effectively extract the features of the polishing area, there are other important information and features that need to be extracted for the entire brake pad. By using a convolutional neural network to extract the feature vector of the entire brake pad image, the overall characteristics and state of the brake pad can be better understood from a global perspective. In an actual production environment, the brake pad may have a variety of different conditions, forms and defects, all of which will affect the results and quality of polishing. By combining the information of the entire brake pad with the information of the polishing area, the robustness and generalization ability of the classifier can be improved, and the false positive rate and missed positive rate can be reduced. In addition to the polishing area and the entire brake pad, there are some other factors that may affect the polishing effect, such as the material of the brake pad, the processing technology, the parameters of the grinder, etc. By fusing the detection feature vector and the polishing feature vector, the influence of multiple factors on the polishing effect can be comprehensively considered, thereby achieving a more accurate and comprehensive classification prediction. Therefore, the detection image is passed through a second convolutional neural network model as a filter to obtain a detection feature vector.

[0048] Furthermore, the grinding feature vector and the detection feature vector are fused to obtain a classification feature vector. Considering that the grinding feature vector and the detection feature vector respectively reflect the different characteristics and information of the grinding area and the entire brake pad, by fusing the feature vectors of the two, the influence of multiple factors on the state of the brake pad can be comprehensively considered, thereby improving the accuracy and stability of the classifier. The grinding feature vector and the detection feature vector are respectively obtained by convolutional neural network processing, and have good expression ability and discrimination. By appropriately fusing the two, the expression and discrimination of the features can be further enhanced. Based on this, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinder stops grinding.

[0049] In particular, when the polishing feature vector and the detection feature vector are fused to obtain the classification feature vector, the polishing feature vector is used to represent the image features of the polishing area of ​​the brake pad, and the detection feature vector is used to represent the global local features of the brake pad. Therefore, the classification feature vector containing the focus feature information of the polishing area and the global local feature information of the brake pad can be obtained by fusing the polishing feature vector and the detection feature vector. However, in the technical solution of the present application, the polishing feature vector is obtained by convolution encoding the polishing region of interest image, and the detection feature vector is obtained by convolution encoding the detection image. From the image domain, the polishing perceptual interest region image is essentially a part of the detection image, and the polishing feature vector and the detection feature vector have the same feature scale, which leads to a large difference in feature density between the polishing feature vector and the detection feature vector, that is, the data change direction and amplitude between each sub-dimension of the polishing feature vector and the detection feature vector are quite different. Under the above premise, if the polishing feature vector and the detection feature vector are fused by calculating the position-weighted sum between the polishing feature vector and the detection feature vector to obtain the classification feature vector, the classification feature vector will become blurred or deviated in the feature local structure, affecting the accuracy of the classification judgment of the classification feature vector.

[0050] Based on this, in the technical solution of the present application, the polishing feature vector and the detection feature vector are subjected to spatial simultaneous projection between each sub-dimension of the feature set to obtain the classification feature vector. Specifically, the covariance matrix between the polishing feature vector and the detection feature vector is first calculated; then, the covariance matrix is ​​subjected to eigenvalue decomposition to obtain multiple eigenvalues ​​and multiple eigenvectors corresponding to the multiple eigenvalues; then, the multiple eigenvectors are arranged as the classification feature vector.

[0051] In this way, the polishing feature vector and the detection feature vector are subjected to spatial joint projection between each sub-dimension based on the feature set, which can fully consider the data change direction and amplitude of the polishing feature vector and the detection feature vector, so that the classification feature vector obtained by spatial joint projection can retain the main feature information of the polishing feature vector and the detection feature vector to the greatest extent and reduce the noise and redundant information in the data. In this way, the patterns and laws in the classification feature vector can be better observed and analyzed. In this way, the accuracy of the classification judgment of the classification feature vector is improved.

[0052] Based on this, the present application provides an intelligent processing and adjustment method for automobile parts, which includes: obtaining a detection image of a polished brake pad; passing the detection image of the polished brake pad through a polishing area target detection network to obtain a polishing area of ​​interest image; passing the polishing area of ​​interest image through a first convolutional neural network model as a filter to obtain a polishing feature vector; passing the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector; performing spatial joint projection between each sub-dimension of a feature set on the polishing feature vector and the detection feature vector to obtain the classification feature vector; and passing the classification feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinder stops grinding.

[0053] Figure 1 FIG. 1 is an application scenario diagram of the intelligent processing and adjustment method for automobile parts according to an embodiment of the present application. Figure 1 As shown, in this application scenario, through a camera (for example, Figure 1 Then, the above image is input to a server (e.g., Figure 1 In S), wherein the server is capable of processing the above-mentioned input image using the automotive parts intelligent processing adjustment algorithm to generate a classification result indicating whether the grinder stops grinding.

[0054] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0055] Exemplary Methods

[0056] Figure 2 FIG. 1 is a flow chart of the intelligent processing and adjustment method for automobile parts according to an embodiment of the present application. Figure 2As shown, according to an embodiment of the present application, it includes: S110, acquiring a detection image of the brake pad after grinding; S120, passing the detection image of the brake pad after grinding through a grinding area target detection network to obtain a grinding area of ​​interest image; S130, passing the grinding area of ​​interest image through a first convolutional neural network model as a filter to obtain a grinding feature vector; S140, passing the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector; S150, performing spatial joint projection between each sub-dimension of a feature set on the grinding feature vector and the detection feature vector to obtain the classification feature vector; and, S160, passing the classification feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinder stops grinding.

[0057] Figure 3 FIG. 1 is a schematic diagram of the structure of the intelligent processing and adjustment method for automobile parts according to an embodiment of the present application. Figure 3 As shown, first, a detection image of the brake pad after grinding is obtained; then, the detection image of the brake pad after grinding is passed through a grinding area target detection network to obtain a grinding area of ​​interest image; then, the grinding area of ​​interest image is passed through a first convolutional neural network model as a filter to obtain a grinding feature vector; at the same time, the detection image is passed through a second convolutional neural network model as a filter to obtain a detection feature vector; then, the grinding feature vector and the detection feature vector are subjected to spatial joint projection between each sub-dimension of a feature set to obtain the classification feature vector; finally, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinder stops grinding.

[0058] In step S110, a test image of the polished brake pad is obtained. By obtaining the test image of the polished brake pad, the area of ​​the brake pad can be quickly and accurately extracted. Through the test image, various information such as the shape, surface defects, and color of the brake pad can be obtained, and this information can be passed to the subsequent processing module to facilitate the comprehensive and in-depth analysis of the brake pad. Through the visual display of the image, the problems of the brake pad can be clearly observed and diagnosed, which helps to improve the efficiency and level of production management and quality control. With the help of advanced image processing technology, the automatic processing and analysis of a large number of brake pad images can be realized, avoiding the errors and instability of manual operation, and improving production efficiency and consistency.

[0059] In step S120, the detection image of the polished brake pad is passed through the polishing area target detection network to obtain a polishing area of ​​interest image. The polishing area in the image can be quickly and accurately detected by the target detection network, which can avoid the complicated process of manually drawing the area of ​​interest, thereby improving processing efficiency and accuracy. For large-scale images and large numbers of images, directly extracting features from the entire image will lead to a sharp increase in the amount of calculation, reducing the response speed and stability of the system. By obtaining the area of ​​interest, the scope and amount of feature extraction can be reduced, and the real-time and reliability of the system can be improved. There may be different areas and features in the polished brake pad. In order to improve the expressiveness and discrimination of the features, it is necessary to extract features for specific polishing areas.

[0060] Specifically, in an embodiment of the present application, the polishing area target detection network is CenterNet, ExtremeNet, RepPoints, YOLOv1 or FCOS.

[0061] In step S130, the image of the polishing area of ​​interest is passed through the first convolutional neural network model as a filter to obtain a polishing feature vector. There may be a variety of features and information in the polishing area, and the most representative and discriminative features need to be extracted to describe the polishing state. The convolutional neural network has the ability to automatically learn feature representation and model training. By taking the image of the polishing area of ​​interest as input, a more expressive and discriminative polishing feature vector can be obtained. By obtaining the polishing feature vector, complex and abstract image information can be converted into a simple and clear numerical vector, which can better reflect the state and characteristics of the polishing area, and help to achieve early warning and management of the grinding machine stopping grinding.

[0062] Figure 4 The flowchart of the method for adjusting the intelligent processing of automobile parts according to the embodiment of the present application is to pass the polishing region of interest image through the first convolutional neural network model as a filter to obtain the polishing feature vector. Figure 4 As shown, the polished image of the region of interest is passed through the first convolutional neural network model as a filter to obtain a polished feature vector, including: using each layer of the first convolutional neural network model as a filter to perform the following on the input data in the forward pass of the layer: S210, convolution processing on the input data to obtain a convolutional feature map; S220, global mean pooling of each feature matrix of the channel dimension of the convolutional feature map to obtain a pooled feature map; and, S230, nonlinear activation of the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the first convolutional neural network as a filter is the polished feature vector, and the input of the first layer of the first convolutional neural network as a filter is the polished image of the region of interest.

[0063] In step S140, the detection image is passed through a second convolutional neural network model as a filter to obtain a detection feature vector. Although the grinding area of ​​interest image can effectively extract the features of the grinding area, there are other important information and features that need to be extracted for the entire brake pad. By using a convolutional neural network to extract the feature vector of the entire brake pad image, the overall characteristics and state of the brake pad can be better understood from a global perspective. In an actual production environment, the brake pad may have a variety of different conditions, forms and defects, all of which will affect the grinding results and quality. By combining the information of the entire brake pad with the information of the grinding area, the robustness and generalization ability of the classifier can be improved, and the false positive rate and missed judgment rate can be reduced. In addition to the grinding area and the entire brake pad, there are some other factors that may affect the grinding effect, such as the material of the brake pad, the processing technology, the parameters of the grinder, etc. By fusing the detection feature vector and the grinding feature vector, the influence of multiple factors on the grinding effect can be comprehensively considered, thereby achieving a more accurate and comprehensive classification prediction.

[0064] Specifically, in an embodiment of the present application, the detection image is passed through a second convolutional neural network model as a filter to obtain a detection feature vector, including: using each layer of the second convolutional neural network model as a filter to perform the following on the input data in the forward pass of the layer: performing convolution processing on the input data to obtain a convolution feature map; performing global mean pooling on each feature matrix of the channel dimension of the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the second convolutional neural network as a filter is the detection feature vector, and the input of the first layer of the second convolutional neural network as a filter is the detection image.

[0065] In step S150, the grinding feature vector and the detection feature vector are subjected to spatial joint projection between the sub-dimensions of the feature set to obtain the classification feature vector. The grinding feature vector and the detection feature vector respectively reflect the different characteristics and information of the grinding area and the entire brake pad. By fusing the feature vectors of the two, the influence of multiple factors on the state of the brake pad can be comprehensively considered to improve the accuracy and stability of the classifier. The grinding feature vector and the detection feature vector are respectively obtained by convolutional neural network processing, and have good expression ability and discrimination. By appropriately fusing the two, the expression and discrimination of the features can be further enhanced. Based on this, the classification feature vector is passed through the classifier to obtain the classification result, and the classification result is used to indicate whether the grinder stops grinding.

[0066] In particular, when the polishing feature vector and the detection feature vector are fused to obtain the classification feature vector, the polishing feature vector is used to represent the image features of the polishing area of ​​the brake pad, and the detection feature vector is used to represent the global local features of the brake pad. Therefore, the classification feature vector containing the focus feature information of the polishing area and the global local feature information of the brake pad can be obtained by fusing the polishing feature vector and the detection feature vector. However, in the technical solution of the present application, the polishing feature vector is obtained by convolution encoding the polishing region of interest image, and the detection feature vector is obtained by convolution encoding the detection image. From the image domain, the polishing perceptual interest region image is essentially a part of the detection image, and the polishing feature vector and the detection feature vector have the same feature scale, which leads to a large difference in feature density between the polishing feature vector and the detection feature vector, that is, the data change direction and amplitude between each sub-dimension of the polishing feature vector and the detection feature vector are quite different. Under the above premise, if the polishing feature vector and the detection feature vector are fused by calculating the position-weighted sum between the polishing feature vector and the detection feature vector to obtain the classification feature vector, the classification feature vector will become blurred or deviated in the feature local structure, affecting the accuracy of the classification judgment of the classification feature vector.

[0067] Specifically, in an embodiment of the present application, the polishing feature vector and the detection feature vector are subjected to spatial joint projection between sub-dimensions of a feature set to obtain the classification feature vector, including: calculating the covariance matrix between the polishing feature vector and the detection feature vector; performing eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues ​​and multiple eigenvectors corresponding to the multiple eigenvalues; and arranging the multiple eigenvectors into the classification feature vector.

[0068] In this way, the polishing feature vector and the detection feature vector are subjected to spatial joint projection between each sub-dimension based on the feature set, which can fully consider the data change direction and amplitude of the polishing feature vector and the detection feature vector, so that the classification feature vector obtained by spatial joint projection can retain the main feature information of the polishing feature vector and the detection feature vector to the greatest extent and reduce the noise and redundant information in the data. In this way, the patterns and laws in the classification feature vector can be better observed and analyzed. In this way, the accuracy of the classification judgment of the classification feature vector is improved.

[0069] In step S160, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinding machine stops grinding.

[0070] Specifically, in an embodiment of the present application, the classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinder stops grinding, including: using multiple fully connected layers of the classifier to fully connect encode the corrected classification feature vector to obtain an encoded classification feature vector; and passing the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0071] In summary, the intelligent processing and adjustment method of automobile parts based on the embodiment of the present application is explained, which uses an artificial intelligence algorithm based on deep learning to obtain the detection image of the brake pad after grinding, and uses a convolutional neural network model to extract the implicit relationship of the grinding area and the implicit relationship of the brake pad as a whole, and further integrates the characteristic information of the two to comprehensively judge whether the grinder should stop grinding. In this way, the intelligent processing and adjustment of automobile parts can be accurately performed to improve the production efficiency and quality level of automobile parts.

[0072] Exemplary Systems

[0073] Figure 5 FIG. 1 is a block diagram of an intelligent processing and adjustment system for automobile parts according to an embodiment of the present application. Figure 5 As shown, according to the embodiment of the present application, the intelligent processing and adjustment system 100 for automobile parts includes: an image acquisition module 110, which is used to acquire a detection image of the brake pad after grinding; a target detection module 120, which is used to pass the detection image of the brake pad after grinding through a grinding area target detection network to obtain a grinding area of ​​interest image; a grinding feature extraction module 130, which is used to pass the grinding area of ​​interest image through a first convolutional neural network model as a filter to obtain a grinding feature vector; a detection feature extraction module 140, which is used to pass the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector; a projection module 150, which is used to perform spatial joint projection between each sub-dimension of a feature set on the grinding feature vector and the detection feature vector to obtain the classification feature vector; and a classification result generation module 160, which is used to pass the classification feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinder stops grinding.

[0074] In one example, in the above-mentioned intelligent processing and adjustment system for automobile parts, the target detection module includes: the polishing area target detection network is CenterNet, ExtremeNet, RepPoints, YOLOv1 or FCOS.

[0075] In one example, in the above-mentioned intelligent processing and adjustment system for automobile parts, the polishing feature extraction module includes: using each layer of the first convolutional neural network model as a filter to perform the following on the input data in the forward pass of the layer: convolution processing on the input data to obtain a convolutional feature map; global mean pooling on each feature matrix of the channel dimension of the convolutional feature map to obtain a pooled feature map; and nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the first convolutional neural network as a filter is the polishing feature vector, and the input of the first layer of the first convolutional neural network as a filter is the polishing area of ​​interest image.

[0076] In one example, in the above-mentioned intelligent processing and adjustment system for automobile parts, the detection feature extraction module includes: using each layer of the second convolutional neural network model as a filter to perform the following on the input data in the forward pass of the layer: convolution processing on the input data to obtain a convolution feature map; global mean pooling on each feature matrix of the channel dimension of the convolution feature map to obtain a pooled feature map; and, nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the output of the last layer of the second convolutional neural network as a filter is the detection feature vector, and the input of the first layer of the second convolutional neural network as a filter is the detection image.

[0077] In one example, in the above-mentioned intelligent processing and adjustment system for automobile parts, the projection module includes: a covariance unit, used to calculate the covariance matrix between the polishing eigenvector and the detection eigenvector; a decomposition unit, used to perform eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues ​​and multiple eigenvectors corresponding to the multiple eigenvalues; and an arrangement unit, used to arrange the multiple eigenvectors into the classification eigenvector.

[0078] In one example, in the above-mentioned intelligent processing and adjustment system for automobile parts, the classification result generation module includes: using multiple fully connected layers of the classifier to fully connect the corrected classification feature vector to obtain an encoded classification feature vector; and passing the encoded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0079] In summary, the intelligent processing and adjustment of automobile parts based on the embodiment of the present application is explained, which uses an artificial intelligence algorithm based on deep learning to obtain the detection image of the brake pad after grinding, and uses a convolutional neural network model to extract the implicit relationship of the grinding area and the implicit relationship of the brake pad as a whole, and further integrates the characteristic information of the two to comprehensively judge whether the grinder should stop grinding. In this way, the intelligent processing and adjustment of automobile parts can be accurately performed to improve the production efficiency and quality level of automobile parts.

[0080] Exemplary Electronic Devices

[0081] Below, reference Figure 6 To describe an electronic device according to an embodiment of the present application. Figure 6 FIG. 1 is a block diagram of an electronic device according to an embodiment of the present application. Figure 6 As shown, the electronic device 10 includes one or more processors 11 and a memory 12 .

[0082] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

[0083] The memory 12 may include one or more computer program products, and the computer program product 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), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may run the program instructions to implement the functions of the intelligent processing and adjustment method for automotive parts of the various embodiments of the present application described above and / or other desired functions. Various contents such as detection images of polished brake pads may also be stored in the computer-readable storage medium.

[0084] In one example, the electronic device 10 may further include: an input device 13 and an output device 14, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0085] The input device 13 may include, for example, a keyboard, a mouse, etc.

[0086] The output device 14 can output various information to the outside, including classification results, etc. The output device 14 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0087] Of course, to simplify, Figure 6 Only some of the components related to the present application in the electronic device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 10 may also include any other appropriate components.

[0088] Exemplary computer program products and computer-readable storage media

[0089] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps in the functions of the intelligent processing and adjustment method of automotive parts according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.

[0090] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0091] In addition, an embodiment of the present application may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps in the functions of the intelligent processing and adjustment method for automotive parts according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0092] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0093] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.

[0094] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0095] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0096] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0097] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. An intelligent processing and adjustment method for automobile parts, characterized in that: include: Acquire a detection image of the polished brake pad; The detection image of the polished brake pad is passed through a polishing area target detection network to obtain a polishing area of ​​interest image; Passing the polished region of interest image through a first convolutional neural network model as a filter to obtain a polished feature vector; Passing the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector; Performing spatial simultaneous projection between sub-dimensions of a feature set on the polishing feature vector and the detection feature vector to obtain a classification feature vector; as well as Passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the grinding machine stops grinding; The step of performing spatial simultaneous projection between sub-dimensions of a feature set on the polishing feature vector and the detection feature vector to obtain the classification feature vector includes: Calculating a covariance matrix between the polishing feature vector and the detection feature vector; Performing eigenvalue decomposition on the covariance matrix to obtain a plurality of eigenvalues ​​and a plurality of eigenvectors corresponding to the plurality of eigenvalues; and Arranging the plurality of feature vectors into the classification feature vector; Among them, the polishing feature vector and the detection feature vector are subjected to spatial joint projection between each sub-dimension based on the feature set, which can fully consider the data change direction and amplitude of the polishing feature vector and the detection feature vector, so that the classification feature vector obtained by the spatial joint projection can retain the main feature information of the polishing feature vector and the detection feature vector to the greatest extent and reduce the noise and redundant information in the data. In this way, the patterns and laws in the classification feature vector can be better observed and analyzed, thereby improving the accuracy of the classification judgment of the classification feature vector.

2. The intelligent processing and adjustment method for automobile parts according to claim 1 is characterized in that: The polishing area target detection network is CenterNet, ExtremeNet, RepPoints, YOLOv1 or FCOS.

3. The intelligent processing and adjustment method for automobile parts according to claim 2 is characterized in that: The polished region of interest image is passed through a first convolutional neural network model as a filter to obtain a polished feature vector, comprising: using each layer of the first convolutional neural network model as a filter to perform the following operations on the input data in the forward pass of the layer: Perform convolution on the input data to obtain a convolution feature map; Performing global mean pooling on each feature matrix of the channel dimension of the convolutional feature map to obtain a pooled feature map; and Performing nonlinear activation on the pooled feature map to obtain an activated feature map; Among them, the output of the last layer of the first convolutional neural network serving as a filter is the polished feature vector, and the input of the first layer of the first convolutional neural network serving as a filter is the polished region of interest image.

4. The intelligent processing and adjustment method for automobile parts according to claim 3 is characterized in that: Passing the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector, comprising: using each layer of the second convolutional neural network model as a filter to perform the following on the input data in the forward pass of the layer: Perform convolution on the input data to obtain a convolution feature map; Performing global mean pooling on each feature matrix of the channel dimension of the convolutional feature map to obtain a pooled feature map; and Performing nonlinear activation on the pooled feature map to obtain an activated feature map; The output of the last layer of the second convolutional neural network as a filter is the detection feature vector, and the input of the first layer of the second convolutional neural network as a filter is the detection image.

5. The intelligent processing and adjustment method for automobile parts according to claim 4 is characterized in that: The classification feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the grinding machine stops grinding, including: Performing full connection encoding on the corrected classification feature vector using multiple fully connected layers of the classifier to obtain an encoded classification feature vector; and The encoded classification feature vector is passed through the Softmax classification function of the classifier to obtain the classification result.

6. An intelligent processing and adjustment system for automobile parts, characterized in that: include: An image acquisition module, used to acquire a detection image of the polished brake pad; A target detection module, used for passing the detection image of the polished brake pad through a polishing area target detection network to obtain a polishing area of ​​interest image; A polishing feature extraction module, used for passing the polishing region of interest image through a first convolutional neural network model as a filter to obtain a polishing feature vector; A detection feature extraction module, used for passing the detection image through a second convolutional neural network model as a filter to obtain a detection feature vector; A projection module, used for performing spatial joint projection between sub-dimensions of a feature set on the polishing feature vector and the detection feature vector to obtain a classification feature vector; A classification result generating module, used for passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used to indicate whether the grinding machine stops grinding; The projection module includes: a covariance unit for calculating the covariance matrix between the polishing feature vector and the detection feature vector; a decomposition unit for performing eigenvalue decomposition on the covariance matrix to obtain a plurality of eigenvalues ​​and a plurality of eigenvectors corresponding to the plurality of eigenvalues; and an arrangement unit for arranging the plurality of eigenvectors into the classification feature vector; Among them, the polishing feature vector and the detection feature vector are subjected to spatial joint projection between each sub-dimension based on the feature set, which can fully consider the data change direction and amplitude of the polishing feature vector and the detection feature vector, so that the classification feature vector obtained by the spatial joint projection can retain the main feature information of the polishing feature vector and the detection feature vector to the greatest extent and reduce the noise and redundant information in the data. In this way, the patterns and laws in the classification feature vector can be better observed and analyzed, thereby improving the accuracy of the classification judgment of the classification feature vector.

7. The intelligent processing and adjustment system for automobile parts according to claim 6 is characterized in that: The target detection module includes: a polishing area target detection network is CenterNet, ExtremeNet, RepPoints, YOLOv1 or FCOS.

8. The intelligent processing and adjustment system for automobile parts according to claim 7 is characterized in that: The polishing feature extraction module is used to: use each layer of the first convolutional neural network model as a filter to perform the following on the input data in the forward pass of the layer: Perform convolution on the input data to obtain a convolution feature map; Performing global mean pooling on each feature matrix of the channel dimension of the convolutional feature map to obtain a pooled feature map; as well as Performing nonlinear activation on the pooled feature map to obtain an activated feature map; Among them, the output of the last layer of the first convolutional neural network serving as a filter is the polished feature vector, and the input of the first layer of the first convolutional neural network serving as a filter is the polished region of interest image.

9. The intelligent processing and adjustment system for automobile parts according to claim 8, characterized in that: The detection feature extraction module is used to: use each layer of the second convolutional neural network model as a filter to perform the following on the input data in the forward pass of the layer: Perform convolution on the input data to obtain a convolution feature map; Performing global mean pooling on each feature matrix of the channel dimension of the convolutional feature map to obtain a pooled feature map; as well as Performing nonlinear activation on the pooled feature map to obtain an activated feature map; The output of the last layer of the second convolutional neural network as a filter is the detection feature vector, and the input of the first layer of the second convolutional neural network as a filter is the detection image.

Images