Motor stator and rotor iron core sheet stamping precision detection system

Abstract

The invention relates to the technical field of motor manufacturing, and discloses a motor stator and rotor iron core sheet stamping precision detection system. The detection system comprises an image acquisition module which is used for acquiring stamping area image data including a lamination contour, a punching position and a surface appearance; and the deformation detection module is used for carrying out dynamic compensation strategy processing on the image data, inputting a three-dimensional reconstruction processing layer to analyze errors and generating deformation characteristic data. A preprocessing unit of the three-dimensional reconstruction processing layer calibrates blocks and balances gray levels, a feature analysis unit carries out joint modeling based on historical data, a geometric matching layer aligns data, a difference quantization layer models morphological differences, and a compensation decision-making layer carries out fusion to generate deformation feature data. The detection method corresponds to the method. The punching precision of the iron chip can be accurately and efficiently detected, detection is optimized through historical data, the method is suitable for the complex iron chip, good universality and expansibility are achieved, and the motor manufacturing quality and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of motor manufacturing, in particular to a stamping accuracy detection system for motor stator and rotor iron core sheets. Background Art

[0002] In the motor manufacturing industry, the stamping accuracy of the stator and rotor laminations plays a decisive role in the motor's performance and quality. With the rapid development of modern industry, motors are increasingly used in various fields, from traditional industrial production equipment to emerging new energy vehicles and smart homes. The requirements for motor performance are constantly increasing, making the stamping accuracy of laminations even more important.

[0003] In the early days of motor manufacturing, due to the relatively limited application scenarios and low performance requirements for motors, lamination punching accuracy inspection primarily relied on manual experience and simple measuring tools. Calipers, micrometers, and other tools were used to measure the lamination dimensions, while visually inspecting the surface quality and punching positions. This method was extremely inefficient, with inspection speeds far behind the pace of production. Furthermore, human error was significant, making it difficult to ensure accurate and consistent inspections. For example, different workers had varying levels of proficiency in the use of measuring tools, resulting in parallax errors in readings, all of which affected the reliability of inspection results.

[0004] With the development of the manufacturing industry, semi-automated inspection equipment has gradually emerged. This type of equipment has improved inspection efficiency to a certain extent, for example, by using simple mechanical positioning devices to assist in measurement, partially automating dimensional measurement. However, its functionality remains relatively limited. It can only measure some basic dimensions of the core sheet, but cannot comprehensively inspect multiple aspects of the core sheet, such as the lamination contour, punching position, and surface topography. Moreover, the inspection accuracy of semi-automated equipment is significantly reduced for core sheets with complex shapes, making it difficult to meet the requirements of high-precision production.

[0005] In recent years, although some advanced detection technologies have begun to be applied to the stamping accuracy detection of iron chips, such as laser measurement technology and three-coordinate measuring machines, these technologies also have obvious limitations. Laser measurement technology is easily interfered by light when detecting iron chips with complex surface morphology or reflective properties, resulting in increased measurement errors. Although the three-coordinate measuring machine has high accuracy, the equipment cost is expensive, the detection process is time-consuming, and the iron chips need to be measured one by one, which cannot meet the requirements of real-time and rapid detection on large-scale production lines. In addition, these existing technologies lack the effective use of historical data when dealing with the stamping accuracy detection of different batches of iron chips. It is difficult to analyze and predict the current detection results based on the historical stamping situation, and it is impossible to timely discover potential problems in the stamping process and make adjustments.

[0006] In actual production, substandard stamping accuracy of the iron core will cause a series of serious problems. If the lamination contour accuracy is insufficient, the air gap between the motor stator and rotor will be uneven, affecting the electromagnetic performance of the motor, reducing the efficiency of the motor and increasing energy consumption. Inaccurate punching position may cause deviations in the iron core during assembly, affecting the overall structural stability of the motor, and even causing the motor to be unable to be assembled normally. Defects in the surface morphology may cause local overheating of the iron core and shorten the service life of the motor. These problems will not only increase the production cost of the motor, but also affect the market competitiveness of the product, restricting the further development of the motor manufacturing industry. Therefore, it is urgent to develop an efficient, accurate and historical data-based stamping accuracy detection system and method for motor stator and rotor iron cores. Summary of the Invention

[0007] The purpose of the present invention is to provide a system for detecting the stamping accuracy of motor stator and rotor iron core sheets to solve the problems raised in the above background technology.

[0008] To achieve the above-mentioned object, the present invention provides the following technical solution: a system for detecting the stamping accuracy of motor stator and rotor iron core sheets, the system comprising: An image acquisition module is configured to acquire image data of the stamping area of the target iron core sheet, wherein the image data of the stamping area includes first block data corresponding to the lamination contour, second block data corresponding to the punching position, and third block data corresponding to the surface topography. The lamination contour data includes reference data retrieved from a cloud database and measured data collected in real time by an industrial camera; A deformation detection module is used to perform dynamic compensation strategy processing on the stamping area image data and input it into a three-dimensional reconstruction processing layer for error analysis, and generate deformation feature data of the target iron core piece according to the output results of the three-dimensional reconstruction processing layer; The three-dimensional reconstruction processing layer includes a preprocessing unit and a feature analysis unit, wherein the preprocessing unit is used to perform block calibration and grayscale equalization on the image transmission data stream, and the feature analysis unit is obtained by joint modeling based on historical benchmark data and historical error data of multiple historical stamping cycles; the feature analysis unit includes a geometric matching layer, a difference quantization layer and a compensation decision layer connected in sequence.

[0009] Preferably, the geometric matching layer is used to perform spatial alignment processing on multiple block data contained in the image transmission data stream to obtain block geometric correlation data; the difference quantization layer is used to model the morphological difference relationship between the block geometric correlation data corresponding to each block data to obtain dynamic difference feature data; the compensation decision layer is used to perform multi-dimensional fusion based on the dynamic difference feature data and the block geometric correlation data to generate deformation feature data.

[0010] Preferably, the modeling of the morphological difference relationship between the block geometric association data corresponding to each block data to obtain dynamic difference feature data includes: A spatial grid partitioning algorithm is used to identify key deformation nodes in the block geometric association data, and a deformation distribution sequence corresponding to each block data is determined according to the stamping parameters corresponding to each key deformation node; The morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data is calculated, and the dynamic difference feature data between the any two block data are determined based on the morphological matching degree.

[0011] Preferably, the calculation of the morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data includes: When the number of nodes in the deformation distribution sequence corresponding to any two block data is inconsistent, virtual node expansion is performed based on the stamping parameters corresponding to the last node in the node with smaller number of nodes, and the morphological matching degree between nodes with the same stamping parameters is calculated based on the expanded data.

[0012] Preferably, the pre-processing unit is specifically used for: Performing standardized segmentation on the geometric information contained in the first block data, the second block data, and the third block data according to a preset grid rule to obtain standardized first block data, standardized second block data, and standardized third block data; The edges of the standardized first block data and the standardized second block data are sharpened using a dynamic grayscale compensation method, and the noise of the standardized third block data is suppressed using a fixed threshold segmentation method to generate first corrected block data, second corrected block data and third corrected block data; wherein the first corrected block data contains the sharpened first stack contour data and the sharpened second stack contour data.

[0013] Preferably, the pre-processing unit is further used for: Calculating an offset coefficient between the sharpened first lamination profile data and the sharpened second lamination profile data in a historical stamping batch; Predicting an expected deformation range of the sharpened second laminate profile data in the current stamping cycle based on the offset coefficient and a reference parameter of the sharpened first laminate profile data in the current stamping cycle; Target stack contour data is generated based on the sharpened second stack contour data and the expected deformation range of the sharpened second stack contour data in the current stamping cycle, and the block data corresponding to the target stack contour data is used as the first corrected block data.

[0014] Preferably, the difference quantization layer specifically includes: a morphological analysis unit, configured to perform geometric topological analysis on each block of data contained in the block geometric association data, so as to extract a corresponding deformation mapping chain from each block of data; The quantization processing unit is used to dynamically compare the deformation mapping chain extracted from each block data with the corresponding block geometry data to generate dynamic difference feature data.

[0015] Preferably, the difference quantization layer further includes: The error correction unit is used to perform redundant fluctuation filtering on the dynamic difference feature data.

[0016] Preferably, the compensation decision layer specifically includes: a multi-channel fusion unit comprising a plurality of sub-decision nodes, each of which is connected to each block in the dynamic difference feature data and the block geometric association data through parameter mapping; a dynamic parameter optimization unit, configured to adjust the parameter mapping by generating an algorithm for a dynamic compensation strategy so as to minimize the deviation between the deformation feature data and a preset accuracy threshold; The priority configuration unit is used to perform error level classification and compensation strategy prediction based on the dynamic difference feature data and the block geometric association data to generate deformation feature data.

[0017] Preferably, the present invention further includes a method for detecting stamping accuracy of motor stator and rotor iron core sheets, the method comprising: Obtaining stamping area image data of the target iron core sheet, the stamping area image data including first block data corresponding to the lamination contour, second block data corresponding to the punching position, and third block data corresponding to the surface topography, the lamination contour data including reference data retrieved from a cloud database and measured data collected in real time by an industrial camera; Performing dynamic compensation strategy processing on the stamping area image data to generate an image transmission data stream; Inputting the image transmission data stream into a three-dimensional reconstruction processing layer for error analysis; The deformation feature data of the target iron chip is generated according to the output result of the three-dimensional reconstruction processing layer; the three-dimensional reconstruction processing layer includes a preprocessing unit and a feature analysis unit, wherein the preprocessing unit is used to perform block calibration and grayscale equalization on the image transmission data stream, and the feature analysis unit is obtained by joint modeling based on historical benchmark data and historical error data of multiple historical stamping cycles; the feature analysis unit includes a geometric matching layer, a difference quantization layer and a compensation decision layer connected in sequence, the geometric matching layer is used to perform spatial alignment processing on multiple block data contained in the image transmission data stream to obtain block geometric association data; the difference quantization layer is used to model the morphological difference relationship between the block geometric association data corresponding to each block data to obtain dynamic difference feature data; the compensation decision layer is used to perform multi-dimensional fusion based on the dynamic difference feature data and the block geometric association data to generate deformation feature data; The modeling of the morphological difference relationship between the block geometric association data corresponding to each block data to obtain dynamic difference feature data includes: A spatial grid partitioning algorithm is used to identify key deformation nodes in the block geometric association data, and a deformation distribution sequence corresponding to each block data is determined according to the stamping parameters corresponding to each key deformation node; Calculating the morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data, and determining the dynamic difference feature data between the any two block data based on the morphological matching degree; calculating the morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data, including: When the number of nodes in the deformation distribution sequence corresponding to any two block data is inconsistent, virtual node expansion is performed based on the stamping parameters corresponding to the last node in the node with smaller number of nodes, and the morphological matching degree between nodes with the same stamping parameters is calculated based on the expanded data.

[0018] Compared with the prior art, the present invention has the following beneficial effects: In terms of detection accuracy, the system uses a unique image acquisition module to obtain multi-dimensional stamping area image data including stacking contours, punching positions and surface morphology. The stacking contour data includes both baseline data from the cloud database and measured data collected in real time by industrial cameras, which provides a rich and accurate information basis for high-precision detection. The pre-processing unit in the three-dimensional reconstruction processing layer performs block calibration and grayscale balance on the image data, effectively eliminating errors and interference that may occur during the image acquisition process. The feature analysis unit is based on the joint modeling of historical baseline data and historical error data of multiple historical stamping cycles, enabling the system to learn the characteristic laws under different stamping conditions. For example, the block data in the image transmission data stream is spatially aligned in the geometric matching layer to ensure that each part of the data is analyzed under the same standard, which greatly improves the accuracy of detection. Compared with traditional detection methods, it can more accurately detect subtle deformations and position deviations of iron chips.

[0019] From the perspective of detection efficiency, the system realizes an automated detection process. The image acquisition module can quickly obtain image data, and the subsequent dynamic compensation strategy processing and error analysis of the three-dimensional reconstruction processing layer are all completed automatically by the system, which greatly shortens the detection time. On large-scale production lines, real-time detection can be achieved, and timely feedback can be given on the stamping accuracy of iron chips. The production process does not need to be interrupted for a long time due to detection, which greatly improves production efficiency, shortens production cycle, and reduces production costs. For example, in the past, manual inspection required complex measurements of iron chips one by one, but this system can quickly perform batch inspections on a batch of iron chips, greatly increasing the number of inspections per unit time.

[0020] In terms of data utilization and analysis, the system fully utilizes historical data. The joint modeling method of the feature analysis unit enables the system to predict potential problems in the current stamping process based on historical stamping data. By analyzing the relationship between historical baseline data and historical error data, the system can adjust the inspection focus and parameters in advance. When a stamping parameter is found to frequently cause specific accuracy issues in historical data, the system will pay more attention to the relevant areas in the current inspection, providing early warning, helping production personnel to adjust the stamping process in a timely manner, avoiding the production of a large number of substandard products, and improving the product qualification rate.

[0021] Furthermore, the system excels at inspecting complex iron cores. For motor stator and rotor iron cores with complex shapes and diverse structures, the system uses a spatial meshing algorithm to identify key deformation nodes, enabling accurate understanding of the deformation characteristics of different parts of the core. Even when the number of nodes in the deformation distribution sequence corresponding to different block data is inconsistent, the system can still accurately calculate the morphological matching degree and generate reliable dynamic difference feature data through virtual node expansion, thereby achieving comprehensive and accurate inspection of the stamping accuracy of complex iron cores, providing strong technical support for motor manufacturers in producing diverse products.

[0022] The detection system and method of the present invention also boast excellent versatility and scalability. Its modular design allows for flexible adjustment and upgrade of each module based on actual production needs. If production processes change or higher requirements for detection accuracy arise, the image acquisition module can be easily updated, or the algorithms in the 3D reconstruction processing layer can be optimized to adapt to different production scenarios. This provides sustainable technical support for the long-term development of the motor manufacturing industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a working principle diagram of the motor stator and rotor lamination stamping accuracy detection system according to the present invention; Figure 2 Schematic diagram of the working principle of iron core stamping data feature analysis and difference quantification; Figure 3 This is a diagram showing the working principle of matching degree calculation when the number of nodes in the deformation distribution sequence is inconsistent; Figure 4 This is a working principle diagram for the standardization and correction of iron chip block data. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] See also Figures 1-4 The present invention provides a motor stator and rotor iron core stamping accuracy detection system, and its specific implementation method is described in detail below.

[0026] The image acquisition module is responsible for acquiring image data of the stamping area of the target iron core. This data includes the first block data corresponding to the lamination contour, the second block data corresponding to the punching position, and the third block data corresponding to the surface topography. The lamination contour data is quite special, consisting of both baseline data retrieved from a cloud database and measured data collected in real time by an industrial camera. The industrial camera is precisely mounted in a suitable position near the stamping equipment to ensure that its shooting angle can fully and clearly capture the stamping area of the iron core, thereby obtaining high-quality image data and providing a reliable foundation for subsequent inspection and analysis.

[0027] The deformation detection module receives image data of the stamping area from the image acquisition module. First, this data is processed using a dynamic compensation strategy. This step aims to optimize the image data and reduce errors caused by various factors (such as lighting changes and equipment vibration). The processed data is then input into the 3D reconstruction processing layer. This layer comprises a preprocessing unit and a feature analysis unit. The preprocessing unit performs block calibration and grayscale equalization on the image transmission data stream, standardizing the image data and facilitating analysis. The feature analysis unit is modeled based on historical baseline data and historical error data from multiple historical stamping cycles. The feature analysis unit comprises a sequentially connected geometric matching layer, a difference quantization layer, and a compensation decision layer. This series of processes ultimately generates deformation feature data for the target lamination based on the output of the 3D reconstruction layer. This allows for precise detection of the stamping accuracy of the motor stator and rotor laminations.

[0028] The present invention will be further described below in conjunction with Examples 1 to 6: Example 1

[0029] This embodiment focuses on the detailed description of the specific working processes of the geometric matching layer, the difference quantization layer, and the compensation decision layer. The geometric matching layer plays a key role in spatial alignment in the entire system. When the image transmission data stream enters the geometric matching layer, the layer will process the multiple block data contained therein, namely the first block data corresponding to the stacking contour, the second block data corresponding to the punching position, and the third block data corresponding to the surface morphology. It will use advanced spatial alignment algorithms to carefully analyze the position and direction information of each block data in space. By accurately calculating the relative position relationship between each block data, they are aligned in space, so that data that may have position deviations can be placed in a unified spatial reference system, thereby obtaining block geometric correlation data. For example, when processing stacking contour data, the geometric matching layer will identify the feature points of the stacking contour at different shooting angles. By matching and adjusting these feature points, the stacking contour data can be accurately aligned to ensure the accuracy of subsequent analysis.

[0030] The main task of the difference quantization layer is to model the morphological difference relationship between the block geometric association data corresponding to each block data. First, the spatial grid division algorithm is used to perform a detailed analysis of the block geometric association data. The algorithm divides the block geometric association data into small grid areas and identifies key deformation nodes in these grid areas. These key deformation nodes are important signs reflecting the morphological changes of the block data. Then, according to the stamping parameters corresponding to each key deformation node, the deformation distribution sequence corresponding to each block data is determined. Stamping parameters include stamping pressure, stamping speed, etc. Different stamping parameters will cause different degrees of deformation of the iron chip. For example, in the block data corresponding to the punching position, by analyzing the stamping pressure parameters corresponding to the key deformation nodes, the deformation trend of the block data at different positions can be determined, and then the deformation distribution sequence can be obtained.

[0031] Computing the morphological matching between nodes with the same stamping parameters in the deformation distribution sequences corresponding to any two data blocks is a key step in the difference quantization layer. In reality, due to various factors, the number of nodes in the deformation distribution sequences corresponding to any two data blocks may be inconsistent. When this occurs, virtual node expansion is performed based on the stamping parameters corresponding to the last node in the sequence with the smaller number of nodes. Virtual node expansion uses a reasonable algorithm to simulate nodes that are close to the actual situation, so that the two sequences have the same number of nodes for subsequent calculations. Next, the morphological matching between nodes with the same stamping parameters is calculated based on the expanded data. By comparing the shape, position, and other characteristics of these nodes, the degree of similarity between them is determined, thereby obtaining the morphological matching. Based on this morphological matching, the dynamic difference feature data between any two data blocks can be determined, providing accurate data support for subsequent compensation decisions.

[0032] The compensation decision layer performs multi-dimensional fusion based on dynamic difference feature data and block-by-block geometric correlation data. It first passes through a multi-channel fusion unit, which consists of multiple sub-decision nodes. Each sub-decision node is connected to a block in the dynamic difference feature data and block-by-block geometric correlation data through parameter mapping. These sub-decision nodes analyze and process the data of their respective connected blocks, extracting data features from different perspectives. Next, the dynamic parameter optimization unit takes action, adjusting the parameter mapping using a dynamic compensation strategy generation algorithm. This adjustment aims to minimize the deviation between the deformation feature data and a preset accuracy threshold. For example, if the currently calculated deformation feature data indicates a large stamping accuracy deviation in a certain area of the iron core sheet, the dynamic parameter optimization unit adjusts the parameter mapping so that subsequent calculations and analyses focus on that area, thereby more accurately assessing and compensating for the deviation. Finally, the priority configuration unit classifies the error levels and predicts compensation strategies based on the dynamic difference feature data and block-by-block geometric correlation data. It will divide the errors into different levels according to the characteristics and degree of deviation of the data, and predict corresponding compensation strategies for different levels of errors, and finally generate accurate deformation feature data to achieve precise detection of the stamping accuracy of the motor stator and rotor iron chips. Example 2

[0033] This embodiment will describe in detail the specific method adopted by the pre-processing unit when processing the first block data, the second block data, and the third block data.

[0034] When the preprocessing unit processes the first block data, the second block data, and the third block data, it first performs standardized segmentation on the geometric information contained therein according to the preset grid rules. The preset grid rules are pre-set based on the common shapes and sizes of the iron chips and the needs of subsequent analysis. For example, for the first block data corresponding to the lamination outline, it is divided into several grid areas of appropriate size according to the shape characteristics of the lamination, and each grid area contains certain geometric information. Through this standardized segmentation, standardized first block data, standardized second block data, and standardized third block data are obtained, making the data more regular and convenient for subsequent processing and analysis.

[0035] After obtaining the standardized data, the pre-processing unit will use different methods to further process different data blocks. For the standardized first block data and the standardized second block data, the dynamic grayscale compensation method is used to sharpen the edges. The dynamic grayscale compensation method automatically adjusts the grayscale value of each pixel in the image according to the grayscale distribution of the image. In the image of the iron chip, the grayscale changes at the edge are usually more obvious, but the edges may not be clear enough due to factors such as lighting. The dynamic grayscale compensation method can enhance the grayscale difference at the edge, making the edges of the stacking contour and the punching position clearer, thereby improving the accuracy of subsequent analysis. For example, when processing the stacking contour data, the originally blurred edges can be clearly displayed after dynamic grayscale compensation to show the boundaries of the stacking, providing a better basis for subsequent size measurement and shape analysis.

[0036] For the standardized third block data, the pre-processing unit uses a fixed threshold segmentation method to suppress noise. In the image data of the surface morphology of the iron chip, there may be some noise points, which will interfere with the accurate judgment of the surface morphology. The fixed threshold segmentation method divides the pixels in the image into two categories according to a pre-set grayscale threshold: one is the pixels with grayscale values higher than the threshold, and the other is the pixels with grayscale values lower than the threshold. In this way, the noise points in the image can be effectively removed and useful surface morphology information can be retained. For example, when processing surface morphology data, after fixed threshold segmentation, the noise points in the image are removed, and the surface texture and defects and other features are more clearly presented, which is conducive to the evaluation of the surface quality of the iron chip.

[0037] After the above processing, the first corrected block data, the second corrected block data, and the third corrected block data are finally generated. The first corrected block data includes the sharpened first laminate outline data and the sharpened second laminate outline data, which provide high-quality basic data for subsequent error analysis and accuracy testing. Example 3

[0038] When processing the lamination profile data, the preprocessing unit calculates the offset coefficient between the sharpened first and second lamination profile data from historical stamping batches. During the historical stamping process, a large amount of lamination profile data is accumulated. The preprocessing unit then performs a detailed analysis of this historical data, selecting the sharpened first and second lamination profile data as the research targets.

[0039] Set up the first In a historical stamping batch, the position coordinates of the first lamination contour data after sharpening in a certain feature direction are , the position coordinates of the second laminated contour data after sharpening in the same feature direction are , the total number of historical stamping batches is .Offset coefficient The calculation formula is: in, It represents the offset coefficient, which reflects the average relative offset of the two sets of lamination profile data in the feature direction during the historical stamping process; It is The position coordinates of the first lamination profile data in a historical stamping batch in a specific feature direction; It is The position coordinates of the second lamination profile data in the same feature direction in a historical stamping batch; Represents the total number of historical stamping batches.

[0040] By comparing the position and shape differences between the two sets of data across different batches, the offset between them is calculated. For example, in certain historical stamping batches, the first lamination profile may have a certain displacement relative to the second lamination profile in a certain direction. By statistically analyzing this displacement data across multiple batches, an offset coefficient can be derived. This offset coefficient reflects the relative change trend between the two sets of lamination profile data during the historical stamping process.

[0041] Based on the obtained offset coefficient and the baseline parameters of the sharpened first lamination profile data during the current stamping cycle, the pre-processing unit can predict the expected deformation range of the sharpened second lamination profile data during the current stamping cycle. The baseline parameters of the current stamping cycle include the setting parameters of the stamping equipment and the characteristics of the raw material.

[0042] Assume that the reference position coordinates of the first lamination contour data after sharpening in the feature direction of the current stamping cycle are , taking into account the offset coefficient and other influencing factors (assuming the comprehensive influencing factor is ), the expected position range of the second lamination contour data after sharpening in the feature direction of the current stamping cycle is predicted to be , which is the expected deformation range. is the reference position coordinate of the first lamination profile in a specific direction in the current cycle; is the offset coefficient calculated previously; It is an influencing factor determined by comprehensively considering multiple factors (such as the stability of the current stamping equipment, changes in raw material characteristics, etc.), and is used to adjust the boundaries of the expected deformation range.

[0043] Combining the offset coefficient with these baseline parameters and using a pre-established prediction model, the predicted range of deformation expected for the second laminate profile under the current stamping conditions can be estimated. For example, if the offset coefficient indicates that the second laminate profile has exhibited a trend of increasing deviation in a certain direction relative to the first laminate profile during past stamping cycles, and the stamping pressure increases during the current stamping cycle, the prediction model can be used to determine that the second laminate profile is likely to further deflect in that direction, thereby determining its expected deformation range.

[0044] Finally, based on the sharpened second lamination contour data and the expected deformation range of the sharpened second lamination contour data in the current stamping cycle, the preprocessing unit generates the target lamination contour data, and uses the block data corresponding to the target lamination contour data as the first correction block data. When generating the target lamination contour data, the actual data of the second lamination contour and the expected deformation range will be fully considered. If the actual second lamination contour data is close to the boundary of the expected deformation range, it will be appropriately adjusted when generating the target lamination contour data to ensure the accuracy and reliability of the data. The block data corresponding to the generated target lamination contour data is used as the first correction block data, which provides an optimized data basis for the subsequent more accurate analysis of the stamping accuracy of the lamination contour. Example 4

[0045] This embodiment describes the working principle and process of the morphological analysis unit and the quantization processing unit in the difference quantization layer. The morphological analysis unit of the difference quantization layer undertakes the important task of in-depth analysis of the block geometric correlation data. When processing each block of data, it first applies the principles and methods of geometric topology to conduct a comprehensive analysis of the geometric figures represented by the data. Taking the block data corresponding to the laminate contour as an example, the morphological analysis unit will carefully study the overall shape of the laminate contour, the connection method between each contour segment, and their relative position relationship in space. Through this in-depth analysis, it is possible to accurately identify the key areas of the laminate contour that are prone to deformation during the stamping process.

[0046] After identifying the key areas, the morphological analysis unit further extracts the deformation features of these areas and organizes them into an ordered deformation mapping chain. For example, at the corners of certain laminate contours, deformation is prone to occur due to the concentration of stamping stress. The morphological analysis unit records the shape change information of these corners at different stamping stages, including angle changes, side length changes, etc., and organizes this information into a mapping chain that can reflect the deformation process of the area according to the time or sequence of stamping steps. In this way, each block of data has a corresponding deformation mapping chain, and these mapping chains become important basic data for the subsequent work of the quantitative processing unit.

[0047] After obtaining the deformation mapping chain corresponding to each block data, the quantization processing unit begins to dynamically compare it with the block geometry data. The block geometry data accurately describes the actual shape, size, and position of the block. The quantization processing unit compares each deformation feature in the deformation mapping chain with the actual geometric parameters of the corresponding position in the block geometry data. For example, for the block data corresponding to the punching position, the quantization processing unit compares the change data of the punching aperture at different stages recorded in the deformation mapping chain with the actual punching aperture data obtained by measurement in the block geometry data.

[0048] During the comparison process, the quantitative processing unit calculates the difference between the two and converts these difference values into dynamic difference feature data based on pre-set rules and algorithms. For example, by calculating indicators such as the percentage of aperture change and the degree of shape deviation, the difference between the deformation mapping chain and the block geometric data is comprehensively measured. These dynamic difference feature data can intuitively reflect the actual morphological changes of the block data during the stamping process, providing a key quantitative basis for the subsequent judgment of the stamping accuracy of the iron chip. Through this dynamic comparison method, the quantitative processing unit can capture tiny deformation differences, thereby improving the accuracy and reliability of the detection system. Example 5

[0049] In the difference quantization layer, the error correction unit's primary responsibility is to optimize the dynamic difference feature data to remove redundant fluctuations and ensure data accuracy and reliability. In actual stamping production environments, the collected dynamic difference feature data often contains unstable fluctuations due to interference from various factors, such as mechanical vibration of the stamping equipment, electromagnetic interference from the surrounding environment, and subtle variations in the raw materials themselves. These fluctuations are not caused by the actual stamping deformation of the core sheet. If left unaddressed, they can seriously affect the accurate assessment of the core sheet's stamping accuracy.

[0050] When operating, the error correction unit first monitors and analyzes dynamic differential feature data in real time. It scans the data point by point, following a specific time interval or data collection sequence. For example, when monitoring the dynamic differential feature data of a laminated profile, the error correction unit observes the changes in the value of each data point and analyzes whether the trend conforms to normal stamping deformation patterns. If a data point's value fluctuates abnormally, significantly deviating from the trend of the preceding and following data points, the error correction unit will flag it as a data point with potential redundant fluctuations.

[0051] Once data points with potential redundant fluctuations are identified, the error correction unit applies a specific algorithm to process them. One common method is a filtering algorithm based on statistical analysis. This algorithm determines whether a data point falls within the normal fluctuation range based on the data's statistical characteristics, such as the mean and standard deviation. For data points that fall outside this range, the error correction unit applies a specific calculation method. For example, a moving average method replaces the abnormal data point with the average of several normal data points before and after it, resulting in a smoother and more stable data trend.

[0052] When processing the dynamic differential feature data of punching positions, if equipment vibration causes a brief, abnormal fluctuation in the data at a particular punching position, the error correction unit applies the aforementioned algorithm to correct this fluctuation. This effectively filters out redundant fluctuations in the dynamic differential feature data, ensuring that the data more accurately reflects the actual stamping deformation of the core. This data, processed by the error correction unit, provides a more reliable basis for subsequent error analysis and compensation decision-making, thereby improving the accuracy and stability of the entire motor stator and rotor core stamping precision detection system. Example 6

[0053] The multi-channel fusion unit in the compensation decision layer contains multiple sub-decision nodes, each of which is connected to each block in the dynamic difference feature data and block geometric association data through parameter mapping. These sub-decision nodes act like multiple independent analyzers, processing data from different perspectives. For example, when processing stack contour data, one sub-decision node may focus on analyzing the overall shape changes of the stack contour, obtaining the related dynamic difference feature data and block geometric association data through parameter mapping, and analyzing from the perspective of overall shape; another sub-decision node may pay more attention to the edge details of the stack contour, also obtaining the corresponding data through parameter mapping, and conducting an in-depth analysis of the changes in the edge part. Through the parallel operation of multiple sub-decision nodes, various feature information of the data can be comprehensively obtained, providing rich data support for subsequent decision-making.

[0054] The dynamic parameter optimization unit functions on the basis of the work of the multi-channel fusion unit. It adjusts the parameter mapping through a dynamic compensation strategy generation algorithm. In the actual stamping process, different stamping conditions and the characteristics of the iron chip will cause different changes in the data. The dynamic parameter optimization unit will continuously adjust the parameter mapping based on the real-time data analysis results. For example, if it is found that a sub-decision node is inaccurate in the analysis of a certain block of data due to unreasonable parameter settings, the dynamic parameter optimization unit will generate an algorithm based on the dynamic compensation strategy, recalculate and adjust the parameter mapping between the sub-decision node and the corresponding block data, so that the sub-decision node can obtain and analyze data more accurately to minimize the deviation between the deformation feature data and the preset accuracy threshold, thereby improving the detection accuracy of the entire system.

[0055] The priority configuration unit classifies errors and predicts compensation strategies based on dynamic difference feature data and block geometry association data. It classifies errors into different levels based on the error conditions reflected in the data. For example, some minor shape deviations may be classified as lower-level errors, while serious dimensional deviations that affect the performance of the iron core are classified as higher-level errors. Based on different error levels, the priority configuration unit predicts the corresponding compensation strategy. For low-level errors, a simple adjustment strategy may be adopted; for high-level errors, more complex compensation measures are required. Finally, the priority configuration unit integrates the results of the multi-channel fusion unit and the dynamic parameter optimization unit to generate accurate deformation feature data, enabling precise detection and compensation decisions for the stamping accuracy of the motor stator and rotor iron cores.

[0056] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0057] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A motor stator and rotor iron core stamping accuracy detection system, characterized in that: include: An image acquisition module is configured to acquire image data of the stamping area of the target iron core sheet, wherein the image data of the stamping area includes first block data corresponding to the lamination contour, second block data corresponding to the punching position, and third block data corresponding to the surface topography. The lamination contour data includes reference data retrieved from a cloud database and measured data collected in real time by an industrial camera; A deformation detection module is used to perform dynamic compensation strategy processing on the stamping area image data and input it into a three-dimensional reconstruction processing layer for error analysis, and generate deformation feature data of the target iron core piece according to the output results of the three-dimensional reconstruction processing layer; The three-dimensional reconstruction processing layer includes a pre-processing unit and a feature analysis unit, wherein the pre-processing unit is used to perform block calibration and grayscale equalization on the image transmission data stream, and the feature analysis unit is obtained by joint modeling based on historical benchmark data and historical error data of multiple historical stamping cycles; The feature analysis unit comprises a geometric matching layer, a difference quantization layer and a compensation decision layer which are connected in sequence.

2. The motor stator and rotor iron core stamping accuracy detection system according to claim 1, characterized in that: The geometric matching layer is used to perform spatial alignment processing on multiple block data contained in the image transmission data stream to obtain block geometric correlation data; the difference quantization layer is used to model the morphological difference relationship between the block geometric correlation data corresponding to each block data to obtain dynamic difference feature data; The compensation decision layer is used to generate deformation feature data by performing multi-dimensional fusion based on dynamic difference feature data and block geometric association data.

3. The motor stator and rotor iron core stamping accuracy detection system according to claim 2, characterized in that: The modeling of the morphological difference relationship between the block geometric association data corresponding to each block data to obtain dynamic difference feature data includes: A spatial grid partitioning algorithm is used to identify key deformation nodes in the block geometric association data, and a deformation distribution sequence corresponding to each block data is determined according to the stamping parameters corresponding to each key deformation node; The morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data is calculated, and the dynamic difference feature data between the any two block data are determined based on the morphological matching degree.

4. The motor stator and rotor lamination stamping accuracy detection system according to claim 3, characterized in that: The calculation of the morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data includes: When the number of nodes in the deformation distribution sequence corresponding to any two block data is inconsistent, virtual node expansion is performed based on the stamping parameters corresponding to the last node in the node with smaller number of nodes, and the morphological matching degree between nodes with the same stamping parameters is calculated based on the expanded data.

5. The motor stator and rotor lamination stamping accuracy detection system according to claim 1, characterized in that: The pre-processing unit is specifically used for: Performing standardized segmentation on the geometric information contained in the first block data, the second block data, and the third block data according to a preset grid rule to obtain standardized first block data, standardized second block data, and standardized third block data; The edges of the standardized first block data and the standardized second block data are sharpened using a dynamic grayscale compensation method, and the noise of the standardized third block data is suppressed using a fixed threshold segmentation method to generate first corrected block data, second corrected block data and third corrected block data; wherein the first corrected block data contains the sharpened first stack contour data and the sharpened second stack contour data.

6. The motor stator and rotor lamination stamping accuracy detection system according to claim 5, characterized in that: The pre-processing unit is further configured to: Calculating an offset coefficient between the sharpened first lamination profile data and the sharpened second lamination profile data in a historical stamping batch; Predicting an expected deformation range of the sharpened second laminate profile data in the current stamping cycle based on the offset coefficient and a reference parameter of the sharpened first laminate profile data in the current stamping cycle; Target stack contour data is generated based on the sharpened second stack contour data and the expected deformation range of the sharpened second stack contour data in the current stamping cycle, and the block data corresponding to the target stack contour data is used as the first corrected block data.

7. The motor stator and rotor lamination stamping accuracy detection system according to claim 2, characterized in that: The difference quantization layer specifically includes: a morphological analysis unit, configured to perform geometric topological analysis on each block of data contained in the block geometric association data, so as to extract a corresponding deformation mapping chain from each block of data; The quantization processing unit is used to dynamically compare the deformation mapping chain extracted from each block data with the corresponding block geometry data to generate dynamic difference feature data.

8. The motor stator and rotor lamination stamping accuracy detection system according to claim 7, characterized in that: The difference quantization layer further includes: The error correction unit is used to perform redundant fluctuation filtering on the dynamic difference feature data.

9. The motor stator and rotor lamination stamping accuracy detection system according to claim 2, characterized in that: The compensation decision-making layer specifically includes: a multi-channel fusion unit comprising a plurality of sub-decision nodes, each of which is connected to each block in the dynamic difference feature data and the block geometric association data through parameter mapping; a dynamic parameter optimization unit, configured to adjust the parameter mapping by generating an algorithm through a dynamic compensation strategy so as to minimize the deviation between the deformation feature data and a preset accuracy threshold; The priority configuration unit is used to perform error level classification and compensation strategy prediction based on the dynamic difference feature data and the block geometric association data to generate deformation feature data.

10. A method for detecting stamping accuracy of motor stator and rotor iron core sheets, characterized in that: include: Obtaining stamping area image data of the target iron core sheet, the stamping area image data including first block data corresponding to the lamination contour, second block data corresponding to the punching position, and third block data corresponding to the surface topography, the lamination contour data including reference data retrieved from a cloud database and measured data collected in real time by an industrial camera; Performing dynamic compensation strategy processing on the stamping area image data to generate an image transmission data stream; Inputting the image transmission data stream into a three-dimensional reconstruction processing layer for error analysis; Generating deformation feature data of the target iron core piece according to the output result of the three-dimensional reconstruction processing layer; The three-dimensional reconstruction processing layer includes a pre-processing unit and a feature analysis unit, wherein the pre-processing unit is used to perform block calibration and grayscale equalization on the image transmission data stream, and the feature analysis unit is obtained by joint modeling based on historical benchmark data and historical error data of multiple historical stamping cycles; the feature analysis unit includes a geometric matching layer, a difference quantization layer and a compensation decision layer connected in sequence, the geometric matching layer is used to perform spatial alignment processing on multiple block data contained in the image transmission data stream to obtain block geometric association data; the difference quantization layer is used to model the morphological difference relationship between the block geometric association data corresponding to each block data to obtain dynamic difference feature data; the compensation decision layer is used to perform multi-dimensional fusion based on the dynamic difference feature data and the block geometric association data to generate deformation feature data; The modeling of the morphological difference relationship between the block geometric association data corresponding to each block data to obtain dynamic difference feature data includes: A spatial grid partitioning algorithm is used to identify key deformation nodes in the block geometric association data, and a deformation distribution sequence corresponding to each block data is determined according to the stamping parameters corresponding to each key deformation node; Calculating the morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data, and determining the dynamic difference feature data between the any two block data based on the morphological matching degree; calculating the morphological matching degree between nodes with the same stamping parameters in the deformation distribution sequence corresponding to any two block data, including: When the number of nodes in the deformation distribution sequence corresponding to any two block data is inconsistent, virtual node expansion is performed based on the stamping parameters corresponding to the last node in the node with smaller number of nodes, and the morphological matching degree between nodes with the same stamping parameters is calculated based on the expanded data.

Images