Motor shell quality detection system and method

Abstract

The invention relates to the technical field of motor shell quality detection, and discloses a motor shell quality detection system, which comprises a data acquisition module, a data analysis module, a defect identification and classification module and a calibration and output module. Through an integrated architecture technology of motor shell quality detection, the core dimension of quality evaluation can be covered: conventional indexes such as size precision and surface defects are covered, stress distribution and service condition simulation elements are also incorporated, and coupling analysis and risk grading of multi-dimensional data are combined, so that the detection result can more accurately reflect the actual service reliability of the motor shell; and a scientific basis is provided for quality control and installation safety.

Description

Technical Field

[0001] This invention relates to the field of motor housing quality inspection technology, and more specifically discloses a motor housing quality inspection system and method. Background Technology

[0002] The motor housing is a key structural component that encloses and protects the core internal components of the motor (such as the stator, rotor, and bearings). Its quality directly determines the motor's mechanical strength, operating accuracy, heat dissipation efficiency, and long-term reliability. Rigorous testing of the motor housing is necessary because under high-speed and high-load operating conditions, any minute dimensional deviation, material defect, or stress concentration may lead to increased vibration, thermal management failure, or even structural fracture, resulting in overall machine failure. Therefore, it is necessary to test the motor housing.

[0003] The prior art patent document CN119715525A discloses a "method and system for surface quality inspection of motor housing die castings". This method involves real-time acquisition of surface image data of the motor housing die castings, followed by preprocessing operations such as noise reduction and contrast enhancement. Then, a surface quality inspection model trained based on an improved YOLO v8 algorithm is used to perform quality inspection on the preprocessed image data, outputting the type, probability, and location information of defects. Based on the inspection results, a preset sorting strategy is executed to sort the motor housing die castings into qualified or defective product areas, and the inspection results are recorded in a database.

[0004] The patent document with authorization announcement number CN117709799B discloses "a sampling inspection system and method for online quality of motor housing", which includes a sampling inspection system. The sampling inspection system includes: an inspection standard module; a sampling method module; a quality inspection module; a data analysis module and a result judgment module. The sampling method module adopts a flexible sampling method, which dynamically adjusts the sampling plan based on real-time production data and historical quality records. By adopting a flexible sampling method, this invention ensures that the algorithm has sufficient flexibility to adapt to changes in the production process and new quality control requirements.

[0005] While existing technologies can achieve a certain degree of optimization in motor housing quality inspection, improving the efficiency and accuracy of die-cast surface inspection through image preprocessing and improved algorithms, reducing human interference, storing inspection results in a database for subsequent management, adapting to changes in the production process through flexible sampling strategies, ensuring inspection accuracy through multiple inspection methods, and classifying and disposing of inspected parts to reduce resource waste and control production costs, current technologies only focus on surface defect detection of die-cast parts. They lack the ability to detect key quality dimensions such as motor housing dimensional deviations and internal stress distribution. Furthermore, the models have insufficient ability to identify new types of defects and have not established a correlation between defects and mechanical properties to achieve accurate risk classification. They also do not involve in-depth detection of stress-strain mechanical properties, nor do they combine traceability technology to achieve full-chain traceability of inspection data. In addition, their adaptability to motor housings of different materials and complex working conditions is poor, resulting in significant limitations in overall inspection dimensions and scenarios. Summary of the Invention

[0006] The present invention mainly provides a motor housing quality inspection system and method, which can solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution, more specifically, a method for detecting the quality of a motor housing, comprising:

[0008] S1. Collect the dimensional parameters, stress and strain data, and surface multispectral image data of the motor housing to be tested;

[0009] S2. Convert the stress and strain data into principal stress data, correct the principal stress data in combination with the size parameters, and then calculate the actual equivalent stress of the motor housing to be tested using the equivalent stress formula.

[0010] S3. Perform fusion preprocessing on multispectral image data, extract image defect features through morphological gradient formula, and complete defect risk classification by combining actual equivalent stress.

[0011] S4. Integrate the dimensional parameter judgment results, equivalent stress judgment results, and defect risk classification results to output a comprehensive quality judgment conclusion.

[0012] Furthermore, in S1, dimensional parameters are acquired by symmetrically distributed line laser sensors, stress and strain data are acquired by miniature strain gauges attached to stress-sensitive points at the motor housing bearing mounting position and housing weld, and multispectral image data are acquired simultaneously by a visible light camera and an infrared thermal imager.

[0013] Furthermore, in S2, a strain-stress conversion model is used to convert stress-strain data into principal stress data; combined with the dimensional parameters of motor housing wall thickness deviation and stop coaxiality deviation, the principal stress data is corrected by a preset deviation correction coefficient; the corrected principal stress data is substituted into the equivalent stress calculation model to obtain the actual equivalent stress.

[0014] Furthermore, in S3, an adaptive threshold algorithm is used to screen the defect features extracted by the morphological gradient formula to determine the defect location and type; and the defect risk classification is completed based on the ratio range of the actual equivalent stress at the defect location to the yield strength of the motor housing material.

[0015] Furthermore, in S4, the size parameter determination is based on the preset motor housing size tolerance threshold range, and the equivalent stress determination is based on the preset material yield strength ratio threshold; the comprehensive determination conclusion is divided into three categories: qualified, high-risk unqualified, and low-risk unqualified; after the determination is completed, a test report containing equipment calibration parameters and test environment data is generated simultaneously.

[0016] According to another aspect of the present invention, a motor housing quality inspection system is provided. The system is based on the above-mentioned motor housing quality inspection method and specifically includes: a data acquisition module, a data analysis module, a defect identification and grading module, and a calibration and output module.

[0017] The data acquisition module obtains multi-dimensional basic inspection data of the motor housing to be inspected; the data analysis module performs layer-by-layer conversion and correction processing on the collected stress and dimensional data to form stress indicators for quality judgment; the defect identification and grading module preprocesses and extracts features from the collected image data, and completes the risk level classification of defects by combining the stress indicators output by the data analysis module; the calibration and output module performs equipment error compensation and inspection information storage on the entire process inspection data, and integrates the output results of multiple modules to form standardized quality judgment conclusions and reports.

[0018] Furthermore, the data acquisition module includes: a size acquisition module, a stress acquisition module, and a defect acquisition module;

[0019] Size acquisition module: Acquires the size parameters of the motor housing to be inspected through symmetrically distributed line laser sensors;

[0020] Stress acquisition module: Collects stress and strain data by attaching miniature strain gauges to stress-sensitive points on the motor housing bearing mounting position and housing weld seam;

[0021] Defect acquisition module: Simultaneously acquires multispectral image data of the motor housing surface using a visible light camera and an infrared thermal imager.

[0022] Furthermore, the data analysis module includes: a principal stress conversion module, a size deviation correction module, and an equivalent stress calculation module;

[0023] Principal stress conversion module: Uses a strain-stress conversion model to convert the collected stress-strain data into principal stress data;

[0024] Dimensional deviation correction module: Based on the dimensional parameters of motor housing wall thickness deviation and stop coaxiality deviation, the principal stress data is corrected by a preset deviation correction coefficient;

[0025] Equivalent stress calculation module: Substitute the corrected principal stress data into the equivalent stress calculation model to obtain the actual equivalent stress of the motor housing to be tested.

[0026] Furthermore, the defect identification and grading module includes: an image fusion processing module, a feature extraction module, and a stress grading module;

[0027] Image fusion processing module: performs fusion preprocessing on multispectral image data to unify the dimensionality and clarity of the image data;

[0028] Feature extraction module: Extracts defect features from the image using morphological gradient formulas, and uses an adaptive threshold algorithm to filter and determine the location and type of defects;

[0029] Stress classification module: Based on the ratio range of the actual equivalent stress at the location of the defect to the yield strength of the motor housing material, the defect risk is classified.

[0030] Furthermore, the calibration and output module includes: an equipment error correction module, a blockchain traceability module, and a result output module;

[0031] Equipment error correction module: Compensates for equipment errors in the entire process of testing data, eliminating the impact of the testing equipment's own errors on the results;

[0032] Blockchain traceability module: Stores the calibration parameters of the testing equipment, the testing environment data, and the testing process data on the blockchain to achieve traceability of testing information;

[0033] Results output module: Integrates dimensional parameter judgment results, equivalent stress judgment results, and defect risk classification results, outputs a comprehensive quality judgment conclusion, and generates a standardized inspection report.

[0034] The beneficial effects of the motor housing quality inspection system and method of this invention are as follows: Through the integrated architecture technology of motor housing quality inspection, it can cover the core dimensions of quality assessment: not only conventional indicators such as dimensional accuracy and surface defects, but also stress distribution and service condition simulation elements. Combined with multi-dimensional data coupling analysis and risk classification, the inspection results more accurately reflect the actual service reliability of the motor housing, providing a scientific basis for quality control and installation safety. Furthermore, multi-modal synchronous acquisition combined with closed-loop calibration traceability realizes the coupling and linkage of various inspection modules and the full-link data collaboration, breaking the barriers of module fragmentation and accuracy attenuation in traditional inspections. A single setup can complete full-dimensional inspection, improving inspection efficiency and ensuring long-term data reliability, thus enhancing the accuracy of defect identification and stress assessment. Simultaneously, the correction coefficient adaptation can flexibly match the inspection of motor housings of different materials and specifications. Relying on the logic of rapid adjustment of the correction coefficient, the system can quickly adapt to various scenarios, improving the applicability and practical application value of this inspection technology. Attached Figure Description

[0035] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0036] Figure 1 This is a schematic diagram of the system framework;

[0037] Figure 2 This is a flowchart illustrating the method. Detailed Implementation

[0038] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0039] According to one aspect of the invention, such as Figures 1-2 As shown, a motor housing quality inspection system and method are provided, including:

[0040] Step 1: Multi-source data acquisition

[0041] Collect dimensional parameters, stress and strain data, and surface multispectral image data of the motor housing to be tested;

[0042] Specifically, dimensional parameters are acquired using symmetrically distributed line laser sensors, stress and strain data are acquired using miniature strain gauges attached to stress-sensitive points on the motor housing bearing mounting position and housing weld, and multispectral image data are acquired simultaneously using a visible light camera and an infrared thermal imager.

[0043] Firstly, the line laser sensor used consists of three symmetrically distributed high-precision devices, which can accurately collect dimensional parameters such as the end face roundness, stop coaxiality, and wall thickness uniformity of the motor housing. Furthermore, the dimensional acquisition module composed of this sensor correlates its own acquisition frequency with the stress data acquisition frequency at a 1:1 ratio, thereby ensuring that the dimensional data and subsequent stress data achieve precise alignment in the spatiotemporal dimensions.

[0044] The micro strain gauges attached to stress-sensitive points can capture the shell strain data in real time under the condition of simulating the actual assembly preload. The stress acquisition module composed of these strain gauges will respond synchronously to the overall acquisition sequence. The obtained strain data can directly provide an accurate data source for subsequent principal stress conversion. Moreover, the placement of the strain gauges accurately covers areas where stress is easy to concentrate, such as bearing mounting positions and shell welds.

[0045] In addition, the device used to acquire multispectral images integrates a visible light camera and an infrared thermal imager, which together constitute a defect acquisition module. The infrared channel can accurately identify areas of abnormal heat conduction caused by pores inside the motor housing, while the visible light channel can clearly capture defects such as surface microcracks. This module performs preliminary synchronization and integration of the dual-channel images. The specific operation steps are as follows: First, the acquisition trigger times of the visible light camera and the infrared thermal imager in the defect acquisition module are precisely aligned using Beidou time synchronization (e.g., the time error is controlled within ±0.1ms), ensuring that the two acquire visible light image frames and infrared thermal image frames of the same area of ​​the motor housing at the same time. Then, the preset device pose calibration parameters of the module are called to map the pixel coordinates of the infrared thermal image to the coordinate system of the visible light image, realizing precise pixel alignment of the dual-channel images in the spatial dimension. Finally, the data of the two-channel images (e.g., including acquisition timestamps, test area identifiers, and device operating parameters) are associated and bound, and the visible light texture information and infrared heat conduction information are encapsulated into associated data groups of the same detection time and the same test area, completing the preliminary synchronization and integration of the dual-channel images and laying the data foundation for subsequent image fusion preprocessing.

[0046] Finally, the entire multi-source data acquisition process relies on BeiDou time synchronization to achieve clock synchronization of multiple devices, controlling the time error of each acquisition device within ±0.1ms. All data from the size acquisition module, stress acquisition module, and defect acquisition module are correlated data from the same detection time, which can effectively avoid subsequent data analysis deviations caused by asynchronous acquisition and ensure the correlation and effectiveness of multi-source data.

[0047] Step 2, Stress-Size Coupling Analysis

[0048] The stress and strain data are converted into principal stress data, and the principal stress data are corrected in combination with the dimensional parameters. Then, the actual equivalent stress of the motor housing to be tested is calculated by the equivalent stress formula.

[0049] Specifically, a strain-stress conversion model is used to convert stress-strain data into principal stress data; combined with the dimensional parameters of motor housing wall thickness deviation and stop coaxiality deviation, the principal stress data is corrected by a preset deviation correction coefficient; the corrected principal stress data is then substituted into the equivalent stress calculation model to obtain the actual equivalent stress.

[0050] First, the principal stress conversion module, responsible for converting stress-strain data into principal stress data, will first retrieve the pre-stored strain-stress conversion model. This model is calibrated based on the elastic modulus, Poisson's ratio, and other mechanical property parameters of the material used in the motor housing. The module will first filter out the strain data transmitted by the stress acquisition module in step 1, removing invalid data caused by strain gauge pasting deviation or instantaneous electromagnetic interference. Then, the filtered valid strain data will be substituted into the conversion model, and the working condition coefficients of the corresponding detection points will be matched. Finally, the first and second principal stress data of each stress-sensitive point will be accurately calculated, and the corresponding detection point identifier and acquisition timestamp will be bound to each set of principal stress data.

[0051] The dimension deviation correction module first receives the core dimension deviation data such as motor housing wall thickness deviation and stop coaxiality deviation transmitted by the dimension acquisition module in step 1. At the same time, it retrieves the pre-stored deviation correction coefficient. This coefficient is calibrated through a large amount of finite element simulation and actual measurement data of motor housings of the same specifications and materials. The module first accurately matches the principal stress data and dimension deviation data according to the detection points to ensure that the principal stress at each point can correspond to its own dimension deviation data. Then, according to the preset correction logic, it combines the wall thickness deviation and stop coaxiality deviation with the corresponding correction coefficient to make targeted corrections to the principal stress data, thereby eliminating the influence of dimension deviation on the actual value of principal stress.

[0052] In addition, during the correction process, the spatiotemporal consistency of the principal stress data and the dimensional deviation data will be verified to ensure that the two sets of data involved in the correction are related data from the same detection time and the same detection location. At the same time, the corresponding correction coefficient will be switched according to the material type of the motor housing to adapt to the detection needs of motor housings of different materials such as aluminum alloy and cast iron, ensuring the accuracy and adaptability of the correction results.

[0053] Finally, the equivalent stress calculation module receives the principal stress data corrected by the size deviation correction module and calls the preset equivalent stress calculation model, which uses the equivalent stress formula of the fourth strength theory:

[0054]

[0055] In the formula, For equivalent stress, and The first and second principal stresses at the detection points are used to substitute the corrected principal stress data into the model to complete the calculation, thereby obtaining the actual equivalent stress at each stress-sensitive point of the motor housing to be tested. The module will also make a preliminary comparison between the calculated actual equivalent stress and the pre-stored material yield strength data to generate a preliminary judgment result on whether the equivalent stress is lower than (e.g., 80%) the material yield strength. At the same time, the actual equivalent stress data, the preliminary judgment result and the corresponding detection point information are integrated to provide accurate stress data support for subsequent defect risk classification.

[0056] Step 3: Defect Identification and Risk Classification

[0057] Multispectral image data is fused and preprocessed, and image defect features are extracted using morphological gradient formulas. Defect risk classification is then completed by combining actual equivalent stress.

[0058] An adaptive threshold algorithm is used to screen the defect features extracted by the morphological gradient formula to determine the defect location and type; the defect risk classification is completed based on the ratio range of the actual equivalent stress at the defect location to the yield strength of the motor housing material.

[0059] First, the module receives the previously synchronized and integrated visible light and infrared image data transmitted from the defect acquisition module. The module will first perform noise reduction and contrast enhancement processing on the two types of images respectively, and deeply fuse the internal heat conduction anomaly information reflected by the infrared channel with the surface texture information of the visible light channel to unify the pixel dimension and visual clarity of the image data. At the same time, it adds detection point and timestamp markers to the fused image to ensure that each image area can accurately correspond to the actual detection part of the motor housing, providing a high-quality image data source for subsequent defect feature extraction.

[0060] Simultaneously, the feature extraction module receives the fused image output by the image fusion processing module, first performs preliminary defect feature extraction on the fused image, and completes the extraction based on the morphological gradient defect feature extraction formula, as shown below:

[0061]

[0062] In the formula, The gradient eigenvalues ​​of the defect. The result of the image dilation operation. The image erosion calculation results are then used to adjust the adaptive threshold according to the actual wall thickness parameters of the motor housing (for example, when the housing wall thickness is <3mm, the threshold will be reduced by 30% to adapt to the micro-defect recognition requirements of thin-walled motor housings). Then, the extracted features are precisely screened based on the adjusted threshold to clearly distinguish the location and specific type of different defects such as microcracks and pores. The corresponding image area coordinates and detection point information are also bound to each identified defect.

[0063] In addition, the stress grading module receives the actual equivalent stress data output by the equivalent stress calculation module in step 2. It first converts the defect location coordinates identified by the feature extraction module into point identifiers consistent with the stress detection points through spatial mapping, thus completing the accurate matching of defect location and stress data. At the same time, it verifies the timestamp information of the two to ensure that they are related data from the same detection time. It also retrieves the corresponding material yield strength benchmark value based on the material information of the motor housing to provide a basis for subsequent risk grading.

[0064] Finally, the defect risk classification is completed based on the ratio range of the actual equivalent stress at the defect location to the material yield strength. For example, if the actual equivalent stress at the defect location exceeds 60% of the material yield strength, the defect is judged as a high-risk defect, and vice versa. Then, the location, type, risk level of the defect and the corresponding stress data are integrated and packaged to generate a complete defect risk classification report, which provides the core defect classification basis for subsequent comprehensive quality assessment.

[0065] Step 4: Quality Judgment and Output

[0066] Integrate the dimensional parameter determination results, equivalent stress determination results, and defect risk classification results to output a comprehensive quality determination conclusion;

[0067] Specifically, the dimensional parameter judgment is based on the preset motor housing dimensional tolerance threshold range, and the equivalent stress judgment is based on the preset material yield strength ratio threshold. The comprehensive judgment conclusion is divided into three categories: qualified, high-risk unqualified, and low-risk unqualified. After the judgment is completed, a test report containing equipment calibration parameters and test environment data is generated simultaneously.

[0068] Firstly, in the core stage of comprehensive judgment, the module receives the original parameters and preliminary judgment results from the dimensional inspection stage. Simultaneously, it retrieves the pre-stored motor housing dimensional tolerance threshold range and conducts secondary verification of dimensional parameters such as end face roundness, stop coaxiality, and wall thickness uniformity to confirm whether all data are within the compliance range. Subsequently, the module connects the actual equivalent stress data and preliminary strength judgment results from the stress coupling analysis stage, matches the yield strength percentage threshold of the corresponding material, and verifies whether the equivalent stress meets the basic requirement of being lower than (e.g., 80%) the material yield strength. In addition, it obtains the complete report from the defect risk classification stage, extracts the specific location, type, and risk level information of the defects, and then accurately associates and matches the three types of data according to the inspection point and timestamp to ensure that the dimensional, stress, and defect data of the same motor housing are associated data from the same inspection batch, thus building a complete and accurate data support system for subsequent comprehensive judgment.

[0069] Meanwhile, during the data verification process, the basic equipment calibration parameters before this detection are retrieved first, covering key data such as the temperature drift correction coefficient of the laser sensor and the zero drift correction coefficient of the strain gauge. Then, the three core detection data of size, stress, and defect are substituted into the preset error compensation model to accurately correct the detection deviations caused by equipment mechanical wear and temperature and humidity fluctuations in the detection environment. After the correction, data consistency verification is also carried out to ensure that the deviation between the corrected data and the original data is within the preset accuracy range, avoiding the interference of equipment self-errors on the accuracy of the final quality determination and providing a highly reliable data source for the comprehensive judgment link;

[0070] Finally, the full-link data evidence preservation work will be initiated. The equipment calibration parameters of this detection, environmental data such as temperature and humidity at the detection time, operator information, original data and corrected data at each link, and preliminary judgment results at each stage are integrated and encapsulated, and the data is uploaded and stored on the blockchain through blockchain technology to achieve the immutability and full-link traceability of the detection data. At the same time, the result output module will complete the final conclusion judgment according to the preset rules. For example, when the size deviation is within the tolerance range, the equivalent stress is lower than 80% of the material yield strength and there are no high-risk defects, it is judged as qualified; if there are high-risk defects, it is judged as high-risk unqualified; if there are only low-risk defects or a single dimension is slightly exceeded, it is judged as low-risk unqualified. Subsequently, a standardized detection report will be generated. The report not only includes the final comprehensive judgment conclusion but also attaches detailed detection data, error correction records, and blockchain evidence preservation query identifiers at each link to complete the display and archiving storage of the report.

[0071] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions, or substitutions made by those of ordinary skill in the art within the essence of the present invention also belong to the protection scope of the present invention.

Claims

1. A method of detecting a mass of a motor housing, characterized by, The method comprises: S1, collecting size parameters, stress and strain data, and surface multispectral image data of the motor shell to be detected; S2, converting the stress and strain data into principal stress data, correcting the principal stress data in combination with the size parameters, and calculating the actual equivalent stress of the motor shell to be detected through an equivalent stress formula; S3, performing fusion preprocessing on the multispectral image data, extracting image defect features through a morphological gradient formula, and completing defect risk grading in combination with the actual equivalent stress; S4, integrating the size parameter determination result, the equivalent stress determination result, and the defect risk grading result, and outputting a comprehensive quality determination conclusion.

2. The method of claim 1, wherein: In S1, the size parameters are collected by symmetrically distributed line laser sensors, the stress and strain data are collected by micro strain gauges pasted on stress sensitive points at bearing mounting positions and shell welds of the motor shell, and the multispectral image data are synchronously collected by a visible light camera and an infrared thermal imager.

3. The method of claim 1, wherein: In S2, a strain-stress conversion model is used to convert the stress and strain data into principal stress data; in combination with the size parameters of the motor shell wall thickness deviation and the joint coaxiality deviation, the principal stress data are corrected through a preset deviation correction coefficient; and the corrected principal stress data are substituted into an equivalent stress calculation model to obtain the actual equivalent stress.

4. The method of claim 1, wherein: In S3, an adaptive threshold algorithm is used to screen the defect features extracted by the morphological gradient formula to determine the defect position and type; and the defect risk grading is completed according to the ratio interval of the actual equivalent stress of the defect position to the yield strength of the motor shell material.

5. The method of claim 1, wherein: In S4, the size parameter determination is based on a preset motor shell size tolerance threshold range, and the equivalent stress determination is based on a preset material yield strength proportion threshold; the comprehensive determination conclusion is divided into three categories: qualified, high-risk unqualified, and low-risk unqualified; After the determination is completed, a detection report containing equipment calibration parameters and detection environment data is synchronously generated.

6. A motor case quality detection system characterized by comprising: The system is realized based on the motor shell quality detection method of any one of claims 1-5, and specifically comprises a data acquisition module, a data analysis module, a defect identification and grading module, and a calibration and output module. The data acquisition module acquires multi-dimensional basic detection data of the motor shell to be detected; the data analysis module performs layer-by-layer conversion and correction processing on the collected stress data and size data to form a stress index for quality determination; the defect identification and grading module performs preprocessing and feature extraction on the collected image data, and completes risk level division of defects in combination with the stress index output by the data analysis module; and the calibration and output module performs equipment error compensation and detection information notarization on the whole-process detection data, and integrates the output results of the multiple modules to form a standardized quality determination conclusion and report.

7. The motor case quality detection system of claim 6, wherein: The data acquisition module comprises a size acquisition module, a stress acquisition module, and a defect acquisition module; The size acquisition module collects size parameters of the motor shell to be detected through symmetrically distributed line laser sensors; The stress acquisition module collects stress and strain data through micro strain gauges pasted on stress sensitive points at bearing mounting positions and shell welds of the motor shell; The defect collection module: through the visible light camera and the infrared thermal imager, the multispectral image data of the motor shell surface is synchronously collected.

8. The motor case quality detection system of claim 6, wherein: The data analysis module comprises a principal stress conversion module, a size deviation correction module and an equivalent stress calculation module. The principal stress conversion module: the strain-stress conversion model is used to convert the collected stress and strain data into principal stress data. The size deviation correction module: combined with the size parameters of the motor shell wall thickness deviation and the joint coaxiality deviation, the principal stress data is corrected through the preset deviation correction coefficient. The equivalent stress calculation module: the corrected principal stress data is substituted into the equivalent stress calculation model to obtain the actual equivalent stress of the motor shell to be detected.

9. The motor case quality detection system of claim 6, wherein: The defect identification and grading module comprises an image fusion processing module, a feature extraction module and a stress grading module. The image fusion processing module: the multispectral image data is fused and preprocessed to unify the image data dimension and the definition. The feature extraction module: the defect features in the image are extracted through the morphological gradient formula, and the adaptive threshold algorithm is used to screen and determine the defect position and type. The stress grading module: according to the ratio interval of the actual equivalent stress of the defect position and the yield strength of the motor shell material, the defect risk grading is completed.

10. The motor case quality detection system of claim 6, wherein: The calibration and output module comprises a device error correction module, a blockchain traceability module and a result output module. The device error correction module: the device error compensation is performed on the whole process detection data to eliminate the influence of the detection device itself error on the result. The blockchain traceability module: the detection device calibration parameters, the detection environment data and the detection process data are chained and stored to realize the traceability of the detection information. The result output module: the size parameter judgment result, the equivalent stress judgment result and the defect risk grading result are integrated to output the comprehensive quality judgment conclusion and generate the standardized detection report.

Images