Intelligent sorting method and sorting system for motor magnetic steel sheets

Abstract

The invention provides an intelligent sorting method and sorting system for motor magnetic steel sheets, and relates to the technical field of motor manufacturing, and the intelligent sorting method comprises the steps that a motor magnetic steel sheet sorting platform is established; performing image capture and acquisition on the target motor magnetic steel sheet through a visual detection camera to obtain appearance image information of the magnetic steel sheet; performing preprocessing and characteristic loss analysis on the appearance image information of the magnetic steel sheet to obtain loss characteristic data of the magnetic steel sheet; building a motor magnetic steel sheet sorting logic device to carry out quality evaluation on the magnetic steel sheet loss characteristic data to obtain a magnetic steel sheet quality sorting result; and based on the quality sorting result of the magnetic steel sheets, sorting execution control is carried out on the target motor magnetic steel sheets. The technical problem of low sorting accuracy caused by lack of comprehensive detection and systematic evaluation on the appearance of the magnetic steel sheet in the prior art is solved, and the technical effect of improving the sorting precision and the sorting efficiency of the magnetic steel sheet of the motor is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of motor manufacturing, and in particular to an intelligent sorting method and sorting system for motor magnetic steel sheets. Background Art

[0002] Motor magnetic steel sheets are important functional materials in motors, and their quality directly affects the performance and life of the motor. During the production process, motor magnetic steel sheets will inevitably have surface defects, dimensional deviations and other problems. Therefore, strict quality sorting of magnetic steel sheets is required during production.

[0003] Existing methods for sorting motor magnetic steel sheets mostly rely on traditional physical measurements or simple manual inspections. In terms of physical measurements, they often focus on measuring the basic dimensions of the magnetic steel sheets and roughly detecting the magnetic strength. Manual inspections rely mainly on the operator's experience and judgment, such as simply checking with the naked eye to see if there are obvious defects in the magnetic steel sheets. However, on the one hand, these existing methods are not comprehensive enough and can only obtain limited information. It is difficult to identify appearance defects such as subtle scratches on the surface of the magnetic steel sheets and tiny irregularities in shape in a timely manner. On the other hand, there is a lack of a systematic evaluation system, and most of them rely on a single indicator or experience judgment, which is highly subjective and lacks accuracy. This greatly affects the sorting quality of the motor magnetic steel sheets, and may in turn affect the overall performance and production efficiency of the motor. Summary of the invention

[0004] The present application provides an intelligent sorting method and sorting system for motor magnetic steel sheets, which solves the technical problem of low sorting accuracy in the prior art due to the lack of comprehensive detection and systematic evaluation of the appearance of the magnetic steel sheets, and achieves the technical effect of improving the sorting accuracy and sorting efficiency of motor magnetic steel sheets.

[0005] In view of the above problems, on the one hand, the present application provides an intelligent sorting method for motor magnetic steel sheets, the method comprising: establishing a motor magnetic steel sheet sorting platform, the motor magnetic steel sheet sorting platform comprising a visual inspection camera, a test fixture and a sorting control module; fixing the target motor magnetic steel sheet on the test fixture, and capturing the image of the target motor magnetic steel sheet through the visual inspection camera to obtain the appearance image information of the magnetic steel sheet; preprocessing and feature loss analysis of the appearance image information of the magnetic steel sheet to obtain magnetic steel sheet loss feature data; building a motor magnetic steel sheet sorting logic, and performing quality assessment on the magnetic steel sheet loss feature data based on the motor magnetic steel sheet sorting logic to obtain a magnetic steel sheet quality sorting result; and performing sorting execution control on the target motor magnetic steel sheet based on the magnetic steel sheet quality sorting result through the sorting control module.

[0006] On the other hand, the present application also provides an intelligent sorting system for motor magnetic steel sheets, the system comprising: a sorting platform unit, used to establish a motor magnetic steel sheet sorting platform, the motor magnetic steel sheet sorting platform comprising a visual inspection camera, a test fixture and a sorting control module; an appearance image capture unit, used to fix the target motor magnetic steel sheet on the test fixture, and to capture and collect images of the target motor magnetic steel sheet through the visual inspection camera to obtain the appearance image information of the magnetic steel sheet; a feature loss analysis unit, used to preprocess and perform feature loss analysis on the appearance image information of the magnetic steel sheet to obtain magnetic steel sheet loss feature data; a quality assessment unit, used to build a motor magnetic steel sheet sorting logic, and to perform quality assessment on the magnetic steel sheet loss feature data based on the motor magnetic steel sheet sorting logic to obtain a magnetic steel sheet quality sorting result; a sorting execution unit, used to perform sorting execution control on the target motor magnetic steel sheet based on the magnetic steel sheet quality sorting result through the sorting control module.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: A motor magnetic steel sheet sorting platform including a visual inspection camera, a test fixture and a sorting control module is established to provide a hardware environment for motor magnetic steel sheet sorting. The target motor magnetic steel sheet is fixedly mounted on the test fixture to ensure the stability and accuracy of the detection process. The image of the target motor magnetic steel sheet is captured and collected by the visual inspection camera to obtain the appearance image information of the magnetic steel sheet, thereby realizing the accurate capture of the appearance of the magnetic steel sheet and providing basic data information for subsequent image analysis and feature extraction. The appearance image information of the magnetic steel sheet is preprocessed and feature loss analysis is performed. Preprocessing can improve image quality, remove interference factors such as noise, and make subsequent analysis more accurate. Feature loss analysis can quantify the loss of the appearance features of the magnetic steel sheet, obtain the loss feature data of the magnetic steel sheet, and provide data support for quality assessment. A motor magnetic steel sheet sorting logic is built, and the quality assessment of the magnetic steel sheet loss feature data is performed based on the motor magnetic steel sheet sorting logic to obtain the magnetic steel sheet quality sorting result. Through the algorithm evaluation of the logic, the automatic judgment of the quality of the magnetic steel sheet is realized, and the scientificity and consistency of the sorting decision are ensured. The sorting control module controls the sorting of the target motor magnetic steel sheets based on the magnetic steel sheet quality sorting results, distinguishes magnetic steel sheets of different quality grades, and ensures that the quality of the magnetic steel sheets entering the next production link meets the requirements.

[0008] In summary, the present application constructs a complete system from magnetic steel sheet image acquisition to quality assessment and then to sorting execution control. The appearance image is acquired by a visual inspection camera, and after preprocessing and feature analysis, the quality is assessed using a special sorting logic device. Finally, the sorting is performed by a sorting control module, thereby achieving comprehensive inspection and systematic evaluation of the appearance of magnetic steel sheets, improving the accuracy and efficiency of motor magnetic steel sheet sorting, reducing manual intervention, ensuring the quality stability of magnetic steel sheets, and thereby improving the overall performance and production efficiency of the motor.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic flow chart of a method for intelligently sorting motor magnetic steel sheets provided in an embodiment of the present application.

[0011] Figure 2 A schematic diagram of the process of obtaining magnetic steel sheet loss characteristic data in an intelligent sorting method for motor magnetic steel sheets provided in an embodiment of the present application.

[0012] Figure 3 A schematic diagram of the process of building a motor magnetic steel sheet sorting logic device in an intelligent sorting method for motor magnetic steel sheets provided in an embodiment of the present application.

[0013] Figure 4 A schematic diagram of the structure of an intelligent sorting system for motor magnetic steel sheets provided in an embodiment of the present application.

[0014] Explanation of the reference numerals: sorting platform unit 10 , appearance image capturing unit 20 , feature loss analysis unit 30 , quality assessment unit 40 , sorting execution unit 50 . DETAILED DESCRIPTION

[0015] The embodiments of the present application provide an intelligent sorting method and sorting system for motor magnetic steel sheets, thereby solving the technical problem of low sorting accuracy in the prior art due to the lack of comprehensive inspection and systematic evaluation of the appearance of magnetic steel sheets, and achieving the technical effect of improving the sorting accuracy and sorting efficiency of motor magnetic steel sheets.

[0016] Embodiment 1, as Figure 1 As shown, an embodiment of the present application provides an intelligent sorting method for motor magnetic steel sheets, the method comprising: Step S1: Establishing a motor magnetic steel sheet sorting platform, wherein the motor magnetic steel sheet sorting platform comprises a visual inspection camera, a test fixture and a sorting control module.

[0017] Specifically, the motor magnetic steel sheet sorting platform is an integrated system for automated sorting of motor magnetic steel sheets, including a visual inspection camera, a test fixture, and a sorting control module. Among them, the visual inspection camera is used to capture the appearance image of the motor magnetic steel sheet. The test fixture is used to fix the motor magnetic steel sheet in a suitable position to ensure that the object is stable and accurately positioned during the inspection process. The sorting control module is a system for controlling the sorting action, and performs specific sorting operations according to the results of the sorting logic.

[0018] According to the shape, size and inspection requirements of the motor magnetic steel sheet, select the appropriate visual inspection camera, test fixture and sorting control module to build the motor magnetic steel sheet sorting platform. For example, if the magnetic steel sheet is small and requires high-precision inspection, select a high-resolution visual inspection camera. Then install the visual inspection camera and test fixture in the appropriate position according to the layout requirements, and electrically connect and communicate the visual inspection camera and test fixture with the sorting control module to ensure that data can be accurately transmitted between components. By building a sorting platform including a visual inspection camera, a test fixture and a sorting control module, the hardware and control foundation is provided for the subsequent magnetic steel sheet fixation, image acquisition, analysis and evaluation, and sorting operations, ensuring that the entire sorting process can be carried out in an orderly and accurate manner.

[0019] Step S2: The target motor magnetic steel sheet is fixedly mounted on the test fixture, and the image of the target motor magnetic steel sheet is captured and collected by the visual inspection camera to obtain the appearance image information of the magnetic steel sheet.

[0020] Specifically, the target motor magnetic steel sheet refers to the magnetic steel sheet to be inspected and sorted during the production process. The target magnetic steel sheet may contain various surface defects, dimensional deviations, etc., and its quality needs to be judged by the sorting system.

[0021] Place the target motor magnetic steel sheet in the test fixture, and according to the shape of the magnetic steel sheet, use the positioning structure of the test fixture (such as slots, positioning pins, etc.) to accurately fix the magnetic steel sheet to prevent movement during image capture. Then, start the visual inspection camera and adjust the camera's focal length, aperture, exposure time and other parameters to obtain a clear and complete image of the magnetic steel sheet's appearance. For example, if there is a reflection problem on the surface of the magnetic steel sheet, the aperture size and exposure time can be adjusted to solve it. The camera converts the optical information on the surface of the magnetic steel sheet into a digital image signal, which contains the appearance image information of the magnetic steel sheet, such as shape, color, surface texture, etc.

[0022] Through image capture and collection, the appearance image information of the target motor magnetic steel sheet can be obtained, which provides the original data for the subsequent analysis of the appearance quality of the magnetic steel sheet and ensures that the subsequent analysis has an accurate data source.

[0023] Step S3: Preprocessing and feature loss analysis of the magnetic steel sheet appearance image information to obtain magnetic steel sheet loss feature data.

[0024] Specifically, first use image processing software or algorithm library (such as OpenCV) to preprocess the appearance image information of the magnetic steel sheet, including image denoising, grayscale, filtering and other operations to improve image quality for subsequent analysis. Perform feature loss analysis on the preprocessed appearance image, and first determine the key features of the appearance of the magnetic steel sheet, such as edge integrity, surface flatness, etc. Then, use an image recognition algorithm (such as an algorithm based on template matching or feature point matching) to compare the collected magnetic steel sheet image with the standard magnetic steel sheet image, calculate the degree of difference in edge shape between the current magnetic steel sheet image and the template image, and quantify these differences into magnetic steel sheet loss feature data. These magnetic steel sheet loss feature data are specific description information of the appearance defect features (such as cracks, breakage, etc.) of the target motor magnetic steel sheet, such as crack length, position coordinates, etc.

[0025] The quality of the appearance image information of the magnetic steel sheet is improved through preprocessing, and the loss feature data of the magnetic steel sheet is obtained through feature loss analysis, which provides key data support for accurately evaluating the quality of the magnetic steel sheet and enables subsequent quality evaluation to be carried out based on objective and quantitative data.

[0026] Step S4: construct a motor magnetic steel sheet sorting logic, perform quality assessment on the magnetic steel sheet loss characteristic data based on the motor magnetic steel sheet sorting logic, and obtain a magnetic steel sheet quality sorting result.

[0027] Specifically, the motor magnetic steel sheet sorting logic is a logical judgment system built according to the characteristics and quality requirements of the magnetic steel sheet, which contains a series of rules, algorithms and models for judging the quality of the magnetic steel sheet according to the magnetic steel sheet loss feature data. The motor magnetic steel sheet sorting logic is built using a machine learning algorithm. Exemplarily, a large amount of magnetic steel sheet loss feature data with known quality results can be collected as a training set, and a classification algorithm (such as support vector machine, neural network, logistic regression, etc.) can be used for training to construct a sorting logic. For example, the edge defect data and surface flatness data in the magnetic steel sheet loss feature data are used as input features, and the magnetic steel sheet quality grade (such as qualified, minor defects, and serious defects) is used as the output label for training. After the magnetic steel sheet loss feature data is obtained, it is input into the constructed sorting logic, and the sorting logic is calculated and judged according to the internal algorithm and model to obtain the magnetic steel sheet quality sorting result. This magnetic steel sheet quality sorting result can be a specific magnetic steel sheet quality grade, or it can be a binary classification result of qualified or unqualified quality.

[0028] By building a sorting logic device and using it to perform quality assessment, an automated and objective assessment of the quality of magnetic steel sheets is achieved. The quality grade of the magnetic steel sheets can be accurately judged according to their actual conditions, providing a decision-making basis for sorting execution control.

[0029] Step S5: The sorting control module performs sorting control on the target motor magnetic steel sheets based on the magnetic steel sheet quality sorting results.

[0030] Specifically, the sorting control module performs the sorting operation based on the received magnetic steel sheet quality sorting results using electrical control equipment (such as solenoid valves, motor drivers, etc.). For example, if the magnetic steel sheet is judged to be qualified, the sorting control module can control the solenoid valve to open the passage to the qualified product storage area to allow the magnetic steel sheet to pass; if the magnetic steel sheet is judged to be defective, the sorting control module controls the solenoid valve to open the passage to the defective product storage area.

[0031] The sorting control module performs accurate sorting operations on the magnetic steel sheets according to the quality sorting results, ensuring that magnetic steel sheets of different quality grades are correctly classified, improving product quality and reducing the possibility of unqualified products entering the next production link.

[0032] Further, such as Figure 2 As shown, step S3 includes: Step S31: performing noise identification and classification on the magnetic steel sheet appearance image information to obtain the noise characteristic information of the magnetic steel sheet image data.

[0033] Step S32: Select a filtering adaptive algorithm according to the noise characteristic information of the magnetic steel sheet image data.

[0034] Step S33: Pre-filter the noise characteristic information of the magnetic steel sheet image data based on the adaptive filtering algorithm to obtain an available magnetic steel sheet appearance image.

[0035] Step S34: performing feature area extraction and feature loss analysis on the available magnetic steel sheet appearance image to obtain the magnetic steel sheet loss feature data.

[0036] Specifically, image noise refers to unnecessary or interfering information that exists in image data. When performing image preprocessing, the image processing algorithm is first used to identify and classify the noise in the magnetic steel sheet appearance image information. For example, by analyzing the distribution of image pixel values, a statistical method is used to determine the noise type. If there are isolated, high-intensity black and white pixels in the image, it may be salt and pepper noise; if the distribution of pixel values ​​shows the characteristics of a Gaussian distribution, it may be Gaussian noise. Some specialized image analysis software, such as Matlab's image processing toolbox, can also be used to analyze the statistical characteristics of the image to obtain the noise characteristic information of the magnetic steel sheet image data. By accurately identifying the noise type and its characteristics in the magnetic steel sheet appearance image, a basis is provided for the subsequent selection of a suitable filtering adaptive algorithm, so that noise can be removed in a targeted manner and image quality can be improved.

[0037] The filter adaptive algorithm is an algorithm that can automatically adjust the filter parameters according to the image noise characteristics. Different noise types require different filtering methods. The adaptive algorithm can automatically select the most appropriate filtering method or adjust the filtering parameters according to the characteristics of the noise to achieve the best filtering effect. According to the noise characteristic information of the magnetic steel sheet image data obtained, the filter adaptive algorithm is used to select the corresponding filtering method and adjust the filtering parameters for different types of noise. For example, for salt and pepper noise, an adaptive algorithm with median filtering as the core is selected. When the noise intensity is large, the window size of the median filter can be adjusted to enhance the filtering effect. If it is Gaussian noise, an adaptive algorithm based on Gaussian filtering may be selected to adjust the kernel function parameters of the Gaussian filter according to the standard deviation of the noise. Through the filter adaptive algorithm, targeted filtering processing can be performed according to the specific noise characteristics of the magnetic steel sheet image, improving the filtering effect while reducing the damage to the useful information of the image.

[0038] According to the selected adaptive filtering algorithm, the noise characteristic information of the magnetic steel sheet image data is used as input to perform filtering operations. For example, if the median filtering adaptive algorithm is used, the algorithm will traverse each pixel in the image, determine a window with the pixel as the center (the window size is adaptively adjusted according to the noise intensity), and then take the median of the pixel value in the window to replace the value of the pixel, thereby removing the salt and pepper noise. For the Gaussian filtering adaptive algorithm, the parameters of the Gaussian kernel function are determined according to the standard deviation of the noise, and then a weighted sum is performed on each pixel in the image to reduce the impact of Gaussian noise. The noise in the magnetic steel sheet appearance image is effectively removed by filtering preprocessing, and a usable magnetic steel sheet appearance image is obtained, which improves the clarity and quality of the image, so that the subsequent feature area extraction and feature loss analysis can be based on more accurate image data.

[0039] Use edge detection, color analysis and other technologies to extract feature areas from the denoised magnetic steel sheet appearance image. Then, perform feature loss analysis on these extracted feature areas, compare them with the standard magnetic steel sheet image features, analyze the feature loss or change of the current magnetic steel sheet image in the feature area, and obtain the magnetic steel sheet loss feature data. For example, if the edge of the standard magnetic steel sheet should be smooth, but the edge of the current magnetic steel sheet image is jagged, these differences are quantified into magnetic steel sheet loss feature data by calculating indicators such as the difference in edge smoothness. Through feature area extraction and feature loss analysis, magnetic steel sheet loss feature data can be accurately obtained. These data provide key quantitative basis for subsequent quality assessment and help to accurately judge the quality status of the magnetic steel sheet.

[0040] Through noise identification and classification, adaptive selection of filtering algorithms, image filtering preprocessing, and feature area extraction and analysis, high-quality processing and comprehensive defect analysis of magnetic steel sheet images are achieved, which greatly improves the image availability and data accuracy, and provides reliable data support for subsequent quality assessment and sorting operations.

[0041] Further, step S34 includes: Step S341: construct a grayscale conversion channel, and perform grayscale conversion on the available magnetic steel sheet appearance image based on the grayscale conversion channel to obtain a grayscale image of the magnetic steel sheet appearance.

[0042] Step S342: Perform edge detection on the grayscale image of the magnetic steel sheet appearance through a Canny edge detector to obtain edge information of the magnetic steel sheet.

[0043] Step S343: performing region segmentation and extraction on the magnetic steel sheet appearance grayscale image according to the magnetic steel sheet edge information to obtain characteristic region information of the magnetic steel sheet.

[0044] Step S344: Perform characteristic loss analysis on the characteristic area information of the magnetic steel sheet to obtain the loss characteristic data of the magnetic steel sheet.

[0045] Specifically, the grayscale conversion channel refers to the calculation method or algorithm structure for converting a color image into a grayscale image. The grayscale image only contains brightness information, not color information. Construct a grayscale conversion channel, which converts the color magnetic steel sheet appearance image into a grayscale image. Grayscale conversion can use the weighted summation method to convert the value of the RGB (red, green, and blue) color space into a single grayscale value. For example, according to the conversion formula Gray=0.299R+0.587G+0.114B, a grayscale conversion channel is constructed, where Gray is the grayscale value, and R, G, and B are the values ​​of the red, green, and blue channels, respectively. For the available magnetic steel sheet appearance image, traverse each pixel in the image, substitute its RGB value into this formula to calculate the grayscale value, and thus obtain the magnetic steel sheet appearance grayscale image. By constructing a grayscale conversion channel and performing grayscale conversion, the originally complex color magnetic steel sheet appearance image is converted into a grayscale image, which reduces the amount of data, while highlighting the contour and structural information of the image, providing a more suitable data basis for subsequent edge detection and region segmentation extraction operations, and improving the efficiency and accuracy of subsequent processing.

[0046] The Canny edge detector is an edge detection algorithm that identifies edges in an image through steps such as Gaussian filtering, gradient calculation, and non-maximum suppression. Edge information refers to the position, direction, and other information of the pixel points in the grayscale image of the magnetic steel sheet that represent the edge of the magnetic steel sheet. When using the Canny edge detector, two thresholds must first be determined, namely the high threshold and the low threshold. These two thresholds are used to distinguish strong edges from weak edges. For example, in the OpenCV library, the cv2.Canny function can be used to implement Canny edge detection. Take the grayscale image of the magnetic steel sheet as input, set appropriate high and low thresholds (these thresholds usually need to be adjusted according to the specific situation of the image, for example, if the image contrast is high, a higher threshold can be set; if the image noise is more, the threshold needs to be appropriately lowered), the function calculates and returns the edge information of the magnetic steel sheet based on the gradient information of the image, usually in the form of a binary image, where the edge pixels are white (value 255) and the non-edge pixels are black (value 0). The edge information provides key contour information for subsequent operations such as determining the shape and size of the magnetic steel sheet and subsequent region segmentation and extraction.

[0047] According to the edge information of the magnetic steel sheet, the grayscale image of the magnetic steel sheet appearance is segmented and extracted, and the grayscale image of the magnetic steel sheet appearance is divided into different regions, which contain different characteristic parts of the magnetic steel sheet, such as the main area of ​​the magnetic steel sheet, the area where special marks or structures may exist, etc. Regional segmentation and extraction can use various image segmentation algorithms, such as threshold-based segmentation, region growing or watershed algorithm. For example, the findContours function in the OpenCV library is used to find the contours in the image according to the edge information, and these contours can be used as the boundaries of regional segmentation. For the grayscale image of the magnetic steel sheet appearance, by traversing these contours, the area surrounded by each contour is determined, and then the relevant information of these areas, such as the area, center of gravity position, average grayscale value, etc. of the area, is extracted to form the characteristic area information of the magnetic steel sheet corresponding to each characteristic area. At the same time, the characteristic area can also be further subdivided or merged according to the specific structure or prior knowledge of the magnetic steel sheet to obtain more accurate characteristic area information. These characteristic area information helps to deeply analyze the various characteristic parts of the magnetic steel sheet, provide more targeted data for the subsequent characteristic loss analysis, and improve the accuracy of the quality analysis of the magnetic steel sheet.

[0048] The characteristic loss analysis is performed on the characteristic area information of the obtained magnetic steel sheet to identify the loss characteristics in the characteristic area and obtain the magnetic steel sheet loss characteristic data. These data can be directly used for quality assessment to help determine whether the magnetic steel sheet meets the quality standards, thereby effectively identifying defective magnetic steel sheets and improving the overall quality control level of the motor magnetic steel sheets.

[0049] Further, step S344 includes: Step S344-1: Build a magnetic steel sheet feature twin network, wherein the magnetic steel sheet feature twin network includes a standard twin sub-network and a target twin sub-network with shared weights.

[0050] Step S344-2: Obtain standard image information of the motor magnetic steel sheet, and extract features of the standard image information of the motor magnetic steel sheet and the characteristic area information of the magnetic steel sheet based on the standard twin network and the target twin network, so as to obtain a standard magnetic steel sheet feature set and a target magnetic steel sheet feature set.

[0051] Step S344-3: Construct a magnetic steel sheet characteristic loss function according to the motor magnetic steel sheet application standard.

[0052] Step S344-4: Perform loss analysis on the standard magnetic steel sheet feature set and the target magnetic steel sheet feature set based on the magnetic steel sheet feature loss function to obtain the magnetic steel sheet loss feature data.

[0053] Specifically, the magnetic steel sheet feature twin network is a neural network structure that is specifically used to analyze and compare the features of magnetic steel sheets. The magnetic steel sheet feature twin network consists of two sub-networks, namely the standard twin sub-network and the target twin sub-network, wherein the standard twin sub-network is used to extract features from the standard image information of the motor magnetic steel sheet, and the target twin sub-network is used to extract features from the feature area information of the magnetic steel sheet (i.e., the feature area information of the magnetic steel sheet to be detected). The two sub-networks share weights, that is, the two sub-networks use the same weight parameters in certain layers (such as convolutional layers, etc.) in the network structure, so that when processing similar tasks (here, feature extraction of the standard magnetic steel sheet image and the feature area of ​​the target magnetic steel sheet), the same feature extraction mode is used to improve the generalization ability and training efficiency of the network. The magnetic steel sheet feature twin network can be built using a deep learning framework (such as PyTorch or TensorFlow). First, define the structure of the network, including the input layer, convolutional layer, pooling layer, fully connected layer, etc. For example, in PyTorch, the nn.Module class can be used to create a custom neural network model. When defining the network layer, the corresponding layers of the standard twin network and the target twin network share weight parameters. For example, for the convolution layer, you can create a convolution layer object, and then use this convolution layer object in both the standard twin network and the target twin network to achieve shared weights. Building a magnetic steel sheet feature twin network provides a unified and effective framework for subsequent feature extraction of standard magnetic steel sheet images and target magnetic steel sheet feature areas. The setting of shared weights enables the feature twin network to better learn the common features of magnetic steel sheets and improve the accuracy and efficiency of feature extraction.

[0054] The standard image information of the motor magnetic steel sheet is the image information representing the appearance features of the qualified magnetic steel sheet that is collected and defined in advance. These images are obtained through strict screening and measurement, and can reflect the various features of the magnetic steel sheet in an ideal state. The standard image information of the motor magnetic steel sheet is read from the pre-established standard image database. Then, the standard image information of the motor magnetic steel sheet is input into the standard twin network, and the characteristic area information of the magnetic steel sheet is input into the target twin network. In a deep learning framework (such as PyTorch), the feature extraction process is implemented by a defined forward propagation function. For example, for the input image, after the convolution layer, pooling layer and other operations of the network, the result output at the fully connected layer is the extracted feature set. After this process, the standard magnetic steel sheet feature set and the target magnetic steel sheet feature set are obtained respectively. Among them, the standard magnetic steel sheet feature set is a set of data obtained by the standard twin network to extract the features of the standard image information of the motor magnetic steel sheet, reflecting the representation of various features of the standard magnetic steel sheet in the twin network. The target magnetic steel sheet feature set is a set of data obtained by extracting the feature area information of the magnetic steel sheet by the target twin network, which reflects the representation of the features of the magnetic steel sheet to be detected in the twin network. These two sets of feature sets provide a data basis for the subsequent construction of the magnetic steel sheet feature loss function and loss analysis, making it possible to compare the feature differences between the standard magnetic steel sheet and the target magnetic steel sheet in the same feature representation space.

[0055] The magnetic steel sheet feature loss function is a function that measures the difference between the standard magnetic steel sheet feature set and the target magnetic steel sheet feature set, and is used to quantify the feature loss. The motor magnetic steel sheet application standard is a qualified standard specification for various features of the magnetic steel sheet (such as shape, size, magnetic properties, etc.) formulated based on the use requirements of the magnetic steel sheet in the actual motor. According to the application standard of the motor magnetic steel sheet, the key features for evaluating the quality of the magnetic steel sheet are determined. For example, if the shape and dimensional accuracy of the magnetic steel sheet are critical to the performance of the motor, then when constructing the magnetic steel sheet feature loss function, the shape and size related features will be considered. The magnetic steel sheet feature loss function is constructed based on the determined key features, which can quantify the difference between the two sets of feature sets. This loss function can be based on a variety of different metrics, such as Euclidean distance, cosine similarity, etc. For example, if the Euclidean distance is used as the loss function, the feature loss is the sum of the squared differences between the two sets of feature vectors. The loss function is defined according to the application standard of motor magnetic steel sheets, so that the characteristic loss function of the magnetic steel sheets can accurately reflect the characteristic loss of the magnetic steel sheets in meeting the actual application requirements, quantitatively evaluate the difference between the standard magnetic steel sheet feature set and the target magnetic steel sheet feature set, and provide a calculation basis for subsequent loss analysis.

[0056] Substitute the standard magnetic steel sheet feature set and the target magnetic steel sheet feature set into the magnetic steel sheet feature loss function for calculation. For example, in Python, the two feature sets are used as the input parameters of the function, and the calculation is performed according to the calculation rules defined by the function. The result is the magnetic steel sheet loss feature data. If the value calculated by the loss function is large, it means that the target magnetic steel sheet has a large difference in features from the standard magnetic steel sheet, and there may be quality problems; if the value is small, it means that the target magnetic steel sheet is closer to the standard magnetic steel sheet in features and has better quality.

[0057] Through loss analysis based on the characteristic loss function of the magnetic steel sheet, the magnetic steel sheet loss characteristic data are obtained. These data can accurately quantify the degree of characteristic loss of the target magnetic steel sheet relative to the standard magnetic steel sheet, and achieve accurate evaluation of the quality of the magnetic steel sheet, thereby improving the overall quality control level of the motor magnetic steel sheet and ensuring that only high-quality magnetic steel sheets are used for motor manufacturing.

[0058] Further, step S344-3 includes: Step 1: Identify the loss type of the magnetic steel sheet based on the motor magnetic steel sheet application standard to obtain a set of characteristic loss types of the magnetic steel sheet.

[0059] Step 2: Analyze the degree of quality influence of each characteristic loss type in the characteristic loss type set of the magnetic steel sheet to obtain a set of characteristic loss type influence factors.

[0060] Step 3: Select the characteristic loss function structure according to the characteristic loss characteristic information of the magnetic steel sheet.

[0061] Step 4: Fit the characteristic loss type influencing factor set based on the characteristic loss function structure to construct the magnetic steel sheet characteristic loss function.

[0062] Specifically, the motor magnetic steel sheet application standard specifies the various requirements that the magnetic steel sheet should meet when it is working normally in the motor, including but not limited to standards in terms of physical size, magnetic strength, surface flatness, etc. According to the various indicators in the motor magnetic steel sheet application standard, the types of non-compliance with the standards of the magnetic steel sheet are determined. For example, if the standard stipulates that the length error range of the magnetic steel sheet is within ±0.1mm, exceeding this range is a type of dimensional loss. For magnetic strength, the motor magnetic steel sheet application standard stipulates a standard value range. Below or above this range is a type of magnetic loss. These possible loss types are organized into a set to form a set of characteristic loss types of magnetic steel sheets. Through the identification of magnetic steel sheet loss types, the various characteristic loss types that may occur in the magnetic steel sheet in meeting the motor application standards are clarified, providing a clear object for the subsequent analysis of the degree of influence of each type on quality.

[0063] For each loss type in the characteristic loss type set of magnetic steel sheets, its degree of influence on the quality is analyzed through experiments, simulations or based on empirical data. For example, for the loss type of surface scratches on magnetic steel sheets, the performance of magnetic steel sheet samples with scratches of different degrees in the motor can be collected, including the changes in performance indicators such as the motor's output power and efficiency. The degree of influence of each loss type is quantified to form a characteristic loss type influence factor set. Each element in the characteristic loss type influence factor set corresponds to a loss type in the characteristic loss type set of magnetic steel sheets, and represents a quantitative value of the degree of influence of the loss type on the quality of the magnetic steel sheets. By obtaining the characteristic loss type influence factor set, the degree of influence of different characteristic loss types on the quality of the magnetic steel sheets can be quantified, which provides a basis for reasonably considering the weight of each loss type when constructing the characteristic loss function of the magnetic steel sheets.

[0064] The characteristic loss characteristic information of the magnetic steel sheet is information that describes the special properties of various characteristic loss types of the magnetic steel sheet. For example, some loss types may affect the quality in a linear relationship, while some may be exponential relationships or more complex relationships. By analyzing the characteristic loss characteristic information of the magnetic steel sheet, if it is found that the effect of the dimensional deviation of the magnetic steel sheet on the quality is roughly linear, that is, the larger the dimensional deviation, the more obvious the quality degradation and the closer it is to the linear law, then you can consider selecting a linear function structure as the basic structure of the characteristic loss function. If the effect of a certain type of loss (such as magnetic non-uniformity) on the quality shows an exponential growth trend with the increase of the non-uniformity, it may be necessary to select an exponential function structure or a composite function structure containing exponential terms. Selecting an appropriate characteristic loss function structure based on the characteristic loss characteristic information of the magnetic steel sheet can ensure that the constructed loss function is more in line with the actual situation of the magnetic steel sheet, thereby more accurately reflecting the relationship between the characteristic loss and quality of the magnetic steel sheet.

[0065] Based on the selected characteristic loss function structure, the characteristic loss type influencing factor set is fitted, the parameters in the characteristic loss function structure are adjusted and optimized, and the final magnetic steel sheet characteristic loss function is constructed to ensure that the loss function can accurately reflect the loss of the magnetic steel sheet. For example, the gradient descent optimization algorithm can be used to adjust the parameters of the loss function to minimize the prediction error. For linear functions, the parameters in the function can be determined by the least squares method.

[0066] The characteristic loss function of the magnetic steel sheet is constructed by fitting the set of characteristic loss type influencing factors based on the characteristic loss function structure. It can comprehensively consider the various characteristic loss types of the magnetic steel sheet and their influence on the quality, accurately quantify the relationship between the characteristic loss and quality of the magnetic steel sheet, and provide an effective mathematical tool for evaluating the quality of the magnetic steel sheet.

[0067] Further, such as Figure 3As shown, in step S4 of the embodiment of the present application, a motor magnetic steel sheet sorting logic is constructed, including: Step S41: construct a sorting model coordinate system, wherein the sorting model coordinate system uses the magnetic steel sheet loss characteristic data as the independent variable coordinate.

[0068] Step S42: Collecting and obtaining a motor magnetic steel sheet sorting data set, wherein the motor magnetic steel sheet sorting data set includes qualified magnetic steel sheet data and unqualified magnetic steel sheet data.

[0069] Step S43: performing regional label division on the sorting model coordinate system based on the motor magnetic steel sheet sorting data set, and constructing a sorting logistic regression line.

[0070] Step S44: Based on the sorting logic regression line, a motor magnetic steel sheet sorting logic is constructed, wherein one side of the sorting logic regression line represents that the magnetic steel sheet has qualified quality, and the other side of the logic regression line represents that the magnetic steel sheet has unqualified quality.

[0071] Specifically, the sorting model coordinate system is a mathematical coordinate system constructed for the sorting of motor magnetic steel sheets, which is used to intuitively represent the relevant data relationship of the magnetic steel sheets. First, determine the dimensions in the loss characteristic data of the magnetic steel sheets and create a corresponding sorting model coordinate system. For example, if there are n different loss characteristic data, create an n-dimensional coordinate system, and each dimension corresponds to a loss characteristic data. In this sorting model coordinate system, the loss characteristic data of the magnetic steel sheets are used as variables of the coordinate axis. For example, if the loss characteristic data of the magnetic steel sheets contains data on multiple aspects such as size deviation and magnetic strength deviation, then these data can be used as different coordinate axes (such as x-axis, y-axis, etc.), so that each magnetic steel sheet has a corresponding coordinate point in this coordinate system, and this coordinate point reflects the loss characteristics of the magnetic steel sheet.

[0072] The motor magnetic steel sheet sorting dataset is a collection of magnetic steel sheet samples with known classifications (qualified or unqualified) for training sorting models. Collect a dataset containing qualified and unqualified magnetic steel sheets. For qualified magnetic steel sheet data, magnetic steel sheets that have passed strict quality inspections can be regularly extracted from the production line, and then various measuring equipment (such as dimension measuring tools, magnetic strength testers, etc.) can be used to measure their relevant data and record them. For unqualified magnetic steel sheet data, data can be obtained from products that are judged to be unqualified, and the corresponding measuring equipment is also used for measurement. Aggregate these qualified and unqualified magnetic steel sheet data to generate a motor magnetic steel sheet sorting dataset. Each set of data in this dataset is labeled as qualified or unqualified, which can comprehensively reflect the different state characteristics of magnetic steel sheets.

[0073] The sorting logistic regression line is a straight line (may be a hyperplane in multi-dimensional cases) that divides the qualified and unqualified magnetic steel sheet areas in the sorting model coordinate system. The qualified and unqualified magnetic steel sheet data points in the motor magnetic steel sheet sorting data set are plotted into the sorting model coordinate system. According to the distribution of the qualified and unqualified magnetic steel sheet data in the motor magnetic steel sheet sorting data set in the sorting model coordinate system, the coordinate system is divided into different areas, and each area is marked as qualified or unqualified. Then, a data analysis method (such as a logistic regression algorithm) is used to find a straight line (or hyperplane) that can best divide the qualified and unqualified magnetic steel sheet data points. This straight line is the sorting logistic regression line.

[0074] When there are new magnetic steel sheet data points (the coordinates in the sorting model coordinate system are determined based on the magnetic steel sheet loss characteristic data), the coordinates are substituted into the equation of the sorting logistic regression line. If the calculation result meets certain conditions, the quality of the magnetic steel sheet is judged to be qualified; otherwise, it is judged to be unqualified. For example, in the two-dimensional case, the equation of the sorting logistic regression line is y=ax+b. If y>ax+b represents qualified, then when the new data point is substituted and meets this inequality, it is judged to be qualified; otherwise, it is unqualified.

[0075] Through the above steps, an automated motor magnetic steel sheet sorting logic device is built, which can quickly and accurately sort and judge the quality of magnetic steel sheets, improve the efficiency and accuracy of magnetic steel sheet quality inspection on the production line, and reduce the errors of manual judgment.

[0076] Furthermore, in step S4, a motor magnetic steel sheet sorting logic is built, and then the following steps are further included: Step S45: verifying the performance of the motor magnetic steel sheet sorting logic according to the model performance evaluation index set to obtain the sorting logic performance parameters.

[0077] Step S46: performing an optimization strategy analysis on the motor magnetic steel sheet sorting logic based on the performance parameters of the sorting logic to obtain a sorting logic optimization strategy.

[0078] Step S47: Optimize and update the parameters of the motor magnetic steel sheet sorting logic according to the sorting logic optimization strategy to obtain the sorting logic optimization parameters.

[0079] Step S48: configuring the model parameters of the motor magnetic steel sheet sorting logic based on the optimization parameters of the sorting logic to obtain the motor magnetic steel sheet optimization sorting logic.

[0080] Specifically, the model performance evaluation index set is a set of standards or indicators used to evaluate the performance of the motor magnetic steel sheet sorting logic, which may include classification model evaluation indicators such as accuracy, recall, and F1 value. The motor magnetic steel sheet sorting logic is tested using a test data set (a set of data different from the training data set, used to evaluate the generalization ability of the model). During the test, the prediction results of the sorting logic for each test data point are recorded and compared with the actual label (i.e., a known qualified or unqualified label). Based on the comparison results, evaluation indicators such as accuracy, recall, and F1 value are calculated as the performance parameters of the sorting logic. The performance parameters of the sorting logic obtained through performance verification can objectively reflect the performance of the sorting logic and provide clear goals and directions for subsequent optimization.

[0081] According to the performance parameters of the sorting logic, analyze the problems and shortcomings of the current sorting logic, and formulate corresponding optimization strategies. If the accuracy is low, it may mean that the model has overfitting or underfitting problems; if the recall rate is low, it may mean that the model has poor recognition ability for certain categories of data. For these problems, corresponding optimization strategies can be formulated. For example, if it is an overfitting problem, you can consider increasing the amount of training data and using regularization methods (such as L1 or L2 regularization) to prevent overfitting; if it is a problem of poor recognition ability for certain categories, you can consider collecting more data of this category for training, or adjusting the model structure to improve the recognition ability of this category.

[0082] According to the sorting logic optimization strategy, adjust the parameters of the sorting logic. For example, if the optimization strategy is to increase the amount of training data, then you can collect more motor magnetic steel sheet data and add this data to the training set to retrain the sorting logic; if you adjust the model structure, such as increasing the number of layers or neurons in the neural network, you need to modify the model definition and retrain the model. During the training process, you can use optimization algorithms (such as stochastic gradient descent) to adjust the parameters of the model so that the model performs better on the new training data.

[0083] The optimization parameters of the sorting logic are applied to the model of the motor magnetic steel sheet sorting logic to obtain the motor magnetic steel sheet optimization sorting logic. The motor magnetic steel sheet optimization sorting logic can sort the motor magnetic steel sheets more accurately in practical applications, further improving the efficiency and accuracy of magnetic steel sheet quality inspection on the production line.

[0084] In summary, the intelligent sorting method for motor magnetic steel sheets provided in the embodiment of the present application has the following technical effects: The embodiment of the present application establishes a motor magnetic steel sheet sorting platform including a visual inspection camera, a test fixture and a sorting control module, which can realize the fixed installation and image capture of the target motor magnetic steel sheet. The appearance image information of the magnetic steel sheet is first subjected to noise identification and classification to obtain noise characteristic information, and then a filter adaptive algorithm is selected for filtering preprocessing to obtain a usable image, which helps to remove interference factors and make subsequent analysis based on more accurate image data. The characteristic area information of the magnetic steel sheet is obtained by constructing grayscale conversion channels, edge detection, region segmentation and extraction, and then by building a magnetic steel sheet feature twin network, feature extraction is performed based on the standard image information of the motor magnetic steel sheet and the characteristic area information of the magnetic steel sheet, and a magnetic steel sheet feature loss function is constructed and loss analysis is performed to obtain the magnetic steel sheet loss feature data. This series of operations gradually derives data that can reflect the key characteristics of the magnetic steel sheet quality from the image data. Then, a sorting model coordinate system is constructed with the magnetic steel sheet loss characteristic data as the independent variable coordinate, and a sorting data set containing qualified and unqualified magnetic steel sheet data is collected. The regional labels are divided to construct a sorting logic regression line and a sorting logic is built, so that the quality of the magnetic steel sheet can be evaluated according to the magnetic steel sheet loss characteristic data, and the quality sorting results are obtained and sorting is performed through the sorting control module. In addition, the performance of the sorting logic is verified and optimized according to the performance evaluation index set to obtain an optimized sorting logic, and the performance of the sorting logic is continuously improved, so that the intelligent sorting method of the entire motor magnetic steel sheet can sort the magnetic steel sheet more accurately and efficiently, improve the accuracy, reliability and automation of magnetic steel sheet sorting, reduce erroneous sorting caused by human factors, and ensure the quality of magnetic steel sheets in motor production, thereby improving the overall performance and reliability of the motor.

[0085] Embodiment 2, as Figure 4 As shown, an embodiment of the present application provides an intelligent sorting system for motor magnetic steel sheets, the system comprising: The sorting platform unit 10 is used to establish a motor magnetic steel sheet sorting platform, and the motor magnetic steel sheet sorting platform includes a visual inspection camera, a test fixture and a sorting control module.

[0086] The appearance image capturing unit 20 is used to fix the target motor magnetic steel sheet on the test fixture, and to capture and collect the image of the target motor magnetic steel sheet through the visual inspection camera to obtain the appearance image information of the magnetic steel sheet.

[0087] The characteristic loss analysis unit 30 is used to pre-process and perform characteristic loss analysis on the appearance image information of the magnetic steel sheet to obtain the magnetic steel sheet loss characteristic data.

[0088] The quality assessment unit 40 is used to build a motor magnetic steel sheet sorting logic, and perform quality assessment on the magnetic steel sheet loss characteristic data based on the motor magnetic steel sheet sorting logic to obtain a magnetic steel sheet quality sorting result.

[0089] The sorting execution unit 50 is used to perform sorting execution control on the target motor magnetic steel sheet based on the magnetic steel sheet quality sorting result through the sorting control module.

[0090] Furthermore, the feature loss analysis unit 30 of the embodiment of the present application is used to perform the following steps: Perform noise identification and classification on the appearance image information of the magnetic steel sheet to obtain the noise characteristic information of the magnetic steel sheet image data; select a filtering adaptive algorithm based on the noise characteristic information of the magnetic steel sheet image data; perform filtering preprocessing on the noise characteristic information of the magnetic steel sheet image data based on the filtering adaptive algorithm to obtain a usable magnetic steel sheet appearance image; perform feature region extraction and feature loss analysis on the usable magnetic steel sheet appearance image to obtain the magnetic steel sheet loss feature data.

[0091] Furthermore, feature region extraction and feature loss analysis are performed on the available magnetic steel sheet appearance image to obtain the magnetic steel sheet loss feature data, and the execution steps include: Construct a grayscale conversion channel, perform grayscale conversion on the available magnetic steel sheet appearance image based on the grayscale conversion channel to obtain a magnetic steel sheet appearance grayscale image; perform edge detection on the magnetic steel sheet appearance grayscale image through a Canny edge detector to obtain magnetic steel sheet edge information; perform region segmentation and extraction on the magnetic steel sheet appearance grayscale image according to the magnetic steel sheet edge information to obtain magnetic steel sheet feature region information; perform feature loss analysis on the magnetic steel sheet feature region information to obtain the magnetic steel sheet loss feature data.

[0092] Further, characteristic loss analysis is performed on the characteristic area information of the magnetic steel sheet to obtain the loss characteristic data of the magnetic steel sheet, and the execution steps include: Build a magnetic steel sheet feature twin network, which includes a standard twin sub-network and a target twin sub-network with shared weights; obtain standard image information of the motor magnetic steel sheet, and extract features of the standard image information of the motor magnetic steel sheet and the feature area information of the magnetic steel sheet based on the standard twin sub-network and the target twin sub-network, respectively, to obtain a standard magnetic steel sheet feature set and a target magnetic steel sheet feature set; construct a magnetic steel sheet feature loss function according to the motor magnetic steel sheet application standard; perform loss analysis on the standard magnetic steel sheet feature set and the target magnetic steel sheet feature set based on the magnetic steel sheet feature loss function, to obtain the magnetic steel sheet loss feature data.

[0093] Furthermore, the execution steps of constructing the magnetic steel sheet characteristic loss function include: Based on the motor magnetic steel sheet application standard, the magnetic steel sheet loss type is identified to obtain a magnetic steel sheet characteristic loss type set; the quality influence degree of each characteristic loss type in the magnetic steel sheet characteristic loss type set is analyzed to obtain a characteristic loss type influence factor set; according to the characteristic loss characteristic information of the magnetic steel sheet, a characteristic loss function structure is selected; based on the characteristic loss function structure, the characteristic loss type influence factor set is fitted to construct the magnetic steel sheet characteristic loss function.

[0094] Furthermore, the quality assessment unit 40 in the embodiment of the present application is used to perform the following steps: Construct a sorting model coordinate system, wherein the sorting model coordinate system uses the magnetic steel sheet loss characteristic data as the independent variable coordinate; collect and obtain a motor magnetic steel sheet sorting data set, wherein the motor magnetic steel sheet sorting data set includes qualified magnetic steel sheet data and unqualified magnetic steel sheet data; perform regional label division on the sorting model coordinate system based on the motor magnetic steel sheet sorting data set, and construct a sorting logistic regression line; based on the sorting logistic regression line, build a motor magnetic steel sheet sorting logic, wherein one side of the sorting logistic regression line represents that the magnetic steel sheet quality is qualified, and the other side of the logistic regression line represents that the magnetic steel sheet quality is unqualified.

[0095] Furthermore, the quality assessment unit 40 in the embodiment of the present application is also used to perform the following steps: The performance of the motor magnetic steel sheet sorting logic is verified according to the model performance evaluation index set to obtain the performance parameters of the sorting logic; based on the performance parameters of the sorting logic, the optimization strategy of the motor magnetic steel sheet sorting logic is analyzed to obtain the optimization strategy of the sorting logic; according to the optimization strategy of the sorting logic, the parameters of the motor magnetic steel sheet sorting logic are optimized and updated to obtain the optimization parameters of the sorting logic; based on the optimization parameters of the sorting logic, the model parameters of the motor magnetic steel sheet sorting logic are configured to obtain the optimized sorting logic of the motor magnetic steel sheet.

[0096] Through the above-mentioned detailed description of the intelligent sorting method of motor magnetic steel sheets in this specification, those skilled in the art can clearly understand the intelligent sorting system of motor magnetic steel sheets in this embodiment. For the system disclosed in the second embodiment, since it corresponds to the method disclosed in the first embodiment and has corresponding functional units and beneficial effects, the relevant parts can be referred to the description of the method part.

[0097] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent sorting method for motor magnetic steel sheets, characterized in that: The method comprises: Establishing a motor magnetic steel sheet sorting platform, the motor magnetic steel sheet sorting platform includes a visual inspection camera, a test fixture and a sorting control module; The target motor magnetic steel sheet is fixedly mounted on the test fixture, and the image of the target motor magnetic steel sheet is captured and collected by the visual inspection camera to obtain the appearance image information of the magnetic steel sheet; Preprocessing and analyzing the characteristic loss of the magnetic steel sheet appearance image information to obtain the magnetic steel sheet loss characteristic data; Building a motor magnetic steel sheet sorting logic device, and performing a quality assessment on the magnetic steel sheet loss characteristic data based on the motor magnetic steel sheet sorting logic device to obtain a magnetic steel sheet quality sorting result; The sorting control module controls the sorting of the target motor magnetic steel sheets based on the magnetic steel sheet quality sorting results.

2. The intelligent sorting method for motor magnetic steel sheets according to claim 1, characterized in that: The preprocessing and feature loss analysis of the magnetic steel sheet appearance image information to obtain magnetic steel sheet loss feature data includes: Performing noise identification and classification on the magnetic steel sheet appearance image information to obtain noise characteristic information of the magnetic steel sheet image data; Selecting a filtering adaptive algorithm according to the noise characteristic information of the magnetic steel sheet image data; Based on the adaptive filtering algorithm, the noise characteristic information of the magnetic steel sheet image data is filtered and preprocessed to obtain a usable magnetic steel sheet appearance image; Feature area extraction and feature loss analysis are performed on the available magnetic steel sheet appearance image to obtain the magnetic steel sheet loss feature data.

3. The intelligent sorting method for motor magnetic steel sheets according to claim 2, characterized in that: The extracting feature regions and analyzing feature loss of the available magnetic steel sheet appearance image to obtain the magnetic steel sheet loss feature data includes: Constructing a grayscale conversion channel, and performing grayscale conversion on the available magnetic steel sheet appearance image based on the grayscale conversion channel to obtain a grayscale image of the magnetic steel sheet appearance; Performing edge detection on the grayscale image of the magnetic steel sheet appearance by using a Canny edge detector to obtain edge information of the magnetic steel sheet; Performing regional segmentation and extraction on the magnetic steel sheet appearance grayscale image according to the magnetic steel sheet edge information to obtain characteristic region information of the magnetic steel sheet; Perform characteristic loss analysis on the characteristic area information of the magnetic steel sheet to obtain the loss characteristic data of the magnetic steel sheet.

4. The intelligent sorting method for motor magnetic steel sheets according to claim 3, characterized in that: The performing characteristic loss analysis on the characteristic area information of the magnetic steel sheet to obtain the loss characteristic data of the magnetic steel sheet includes: Building a magnetic steel sheet feature twin network, wherein the magnetic steel sheet feature twin network includes a standard twin sub-network and a target twin sub-network that share weights; Acquire standard image information of the motor magnetic steel sheet, and extract features of the standard image information of the motor magnetic steel sheet and the characteristic region information of the magnetic steel sheet based on the standard twin network and the target twin network, respectively, to obtain a standard magnetic steel sheet feature set and a target magnetic steel sheet feature set; According to the application standard of motor magnetic steel sheets, the characteristic loss function of magnetic steel sheets is constructed; Based on the magnetic steel sheet feature loss function, loss analysis is performed on the standard magnetic steel sheet feature set and the target magnetic steel sheet feature set to obtain the magnetic steel sheet loss feature data.

5. The intelligent sorting method of motor magnetic steel sheets according to claim 4, characterized in that: The method of constructing a magnetic steel sheet characteristic loss function according to the motor magnetic steel sheet application standard includes: Based on the motor magnetic steel sheet application standard, the magnetic steel sheet loss type is identified to obtain a magnetic steel sheet characteristic loss type set; Performing a quality influence analysis on each characteristic loss type in the characteristic loss type set of the magnetic steel sheet to obtain a characteristic loss type influence factor set; Selecting a characteristic loss function structure according to characteristic loss characteristic information of the magnetic steel sheet; The characteristic loss type influencing factor set is fitted based on the characteristic loss function structure to construct the magnetic steel sheet characteristic loss function.

6. The intelligent separation method of motor magnetic steel sheets according to claim 1, characterized in that: The motor magnetic steel sheet sorting logic device is constructed, including: Constructing a sorting model coordinate system, wherein the sorting model coordinate system uses the magnetic steel sheet loss characteristic data as an independent variable coordinate; Acquiring a motor magnetic steel sheet sorting data set, wherein the motor magnetic steel sheet sorting data set includes qualified magnetic steel sheet data and unqualified magnetic steel sheet data; Based on the motor magnetic steel sheet sorting data set, the sorting model coordinate system is divided into regional labels to construct a sorting logistic regression line; Based on the sorting logic regression line, a motor magnetic steel sheet sorting logic is constructed, wherein one side of the sorting logic regression line represents that the magnetic steel sheet has qualified quality, and the other side of the logic regression line represents that the magnetic steel sheet has unqualified quality.

7. The intelligent separation method of motor magnetic steel sheets according to claim 6, characterized in that: The method comprises: According to the model performance evaluation index set, the performance of the motor magnetic steel sheet sorting logic is verified to obtain the performance parameters of the sorting logic; Performing an optimization strategy analysis on the motor magnetic steel sheet sorting logic based on the performance parameters of the sorting logic to obtain an optimization strategy for the sorting logic; According to the sorting logic optimization strategy, the parameters of the motor magnetic steel sheet sorting logic are optimized and updated to obtain the sorting logic optimization parameters; The model parameters of the motor magnetic steel sheet sorting logic are configured based on the optimization parameters of the sorting logic to obtain the motor magnetic steel sheet optimization sorting logic.

8. An intelligent sorting system for motor magnetic steel sheets, characterized in that: The system is used to execute the intelligent sorting method of motor magnetic steel sheets according to any one of claims 1 to 7, comprising: A sorting platform unit, used to establish a motor magnetic steel sheet sorting platform, wherein the motor magnetic steel sheet sorting platform includes a visual inspection camera, a test fixture and a sorting control module; An appearance image capturing unit, used to fix the target motor magnetic steel sheet on the test fixture, and to capture and collect images of the target motor magnetic steel sheet through the visual inspection camera to obtain appearance image information of the magnetic steel sheet; A characteristic loss analysis unit, used for preprocessing the appearance image information of the magnetic steel sheet and performing characteristic loss analysis to obtain magnetic steel sheet loss characteristic data; A quality assessment unit, used to build a motor magnetic steel sheet sorting logic, and to perform quality assessment on the magnetic steel sheet loss characteristic data based on the motor magnetic steel sheet sorting logic to obtain a magnetic steel sheet quality sorting result; The sorting execution unit is used to perform sorting execution control on the target motor magnetic steel sheet based on the magnetic steel sheet quality sorting result through the sorting control module.

Images