A method, apparatus and electronic device for testing drive motor assembly

Abstract

This application discloses a method, apparatus, and electronic device for testing a drive motor assembly. The method includes: establishing a first multi-dimensional testing dataset for the drive motor assembly; identifying process interference features and masking invalid data in the first multi-dimensional testing dataset to obtain a second multi-dimensional testing dataset; then performing cross-validation correction on the second multi-dimensional testing dataset using a multi-parameter coupling judgment model to obtain a calibrated third multi-dimensional valid testing dataset; obtaining a fourth multi-dimensional testing dataset after deviation data compensation based on the coupling accuracy value of the drive motor assembly; and substituting the fourth multi-dimensional testing dataset into a two-dimensional feature matching model to calculate the coupling accuracy value and generate a testing report to achieve the testing of the drive motor assembly. The drive motor assembly testing apparatus and electronic device are adapted to the above-mentioned drive motor assembly testing method and can improve the accuracy of determining the overall compatibility of the motor assembly.

Description

Technical Field

[0001] This application relates to the field of motor testing, and more particularly to a method, apparatus and electronic equipment for testing drive motor assemblies. Background Technology

[0002] The new energy vehicle industry is rapidly evolving, and the drive motor assembly, as a core power component, directly impacts the safety and reliability of the entire vehicle through its assembly precision and overall performance. The industry places stringent demands on its testing accuracy, process standardization, and quality traceability. Traditional drive motor assembly testing and training equipment largely relies on dedicated physical hardware such as insulation testers, airtightness testers, and geometric tolerance measuring instruments. This not only incurs high equipment procurement and maintenance costs, but also exposes the vehicle to tooling interference and measurement errors due to contact-based measurements, limiting testing scenarios. Furthermore, existing testing processes are often segmented, lacking refined data processing after raw data collection. Process interference and abnormal data can easily distort test results. Additionally, the reliance on manual experience to determine pass / fail status leads to inconsistent standards, strong subjectivity, and difficulty in avoiding judgment errors.

[0003] Furthermore, traditional testing can only output a pass or fail conclusion, and cannot trace the root cause of the deviation process for non-conforming samples, or provide a corresponding closed-loop correction process. The rework efficiency of non-conforming products is low, similar defects are easy to recur, the test data is scattered and it is difficult to form a complete traceable report, which cannot meet the needs of large-scale production and full-process quality control.

[0004] In view of the current situation, there is an urgent need for a closed-loop testing method for drive motor assemblies to make up for the shortcomings of existing technologies and meet the testing requirements of high precision, low cost and standardization. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this application is to provide a method, apparatus, and electronic device for testing drive motor assemblies, which can improve the accuracy of determining the overall compatibility of motor assemblies.

[0006] In a first aspect, embodiments of this application provide a method for detecting a drive motor assembly, the method comprising: Obtain geometric and performance characteristic data of the drive motor assembly; based on the virtual parameter fitting algorithm that dynamically associates the assembly and adjustment process and parameters, integrate the actual assembly and adjustment process working condition information to perform standard range matching, interpolation completion, deviation calibration and range constraint fitting on the geometric and performance characteristic data, and establish the first multi-dimensional detection dataset. The first multi-dimensional detection dataset is subjected to temporal identification of process interference features and invalid data masking. Interference distortion and logical conflict data in different assembly and adjustment process stages are distinguished. After dimension alignment and format normalization, the second multi-dimensional detection dataset is obtained. Then, the process dimension cross-validation correction is performed on it through a multi-parameter coupled judgment model to filter out abnormal parameters that do not match the logic of the assembly and adjustment process and complete global deviation calibration to obtain the third multi-dimensional effective detection dataset. A two-dimensional feature matching model is constructed based on the third multi-dimensional effective detection dataset. The coupling sensitivity coefficients of geometric and performance parameters are introduced, and the dataset is substituted to perform coupling analysis and calculate the coupling accuracy value. The coupling accuracy value is compared with a preset threshold. If the threshold is met, a basic detection report is generated. If the threshold is not met, deviation data is extracted from the third multi-dimensional effective detection dataset, the deviation magnitude of each deviation data is calculated, and the coupling influence weight corresponding to each deviation data is calculated and output based on the coupling sensitivity coefficient. Based on deviation data, deviation magnitude, coupling influence weight and preset process traceability rules, the causes of core assembly and adjustment processes and calibration parameters are reversed, the assembly and adjustment process of the motor assembly is simulated and iterated to correct, the data with remaining deviations are coupled and compensated to correct, and the fourth multi-dimensional detection dataset is obtained. The fourth multi-dimensional detection dataset is back-substituted into the model to recalculate the coupling accuracy value. The effectiveness of the correction is verified based on the change in accuracy value between the two tests. The detection report is generated by integrating the full-process detection data, the causes of deviation, the correction measures, and the verification results, thus realizing the detection of the drive motor assembly.

[0007] In one possible implementation, geometric and performance characteristic data of the drive motor assembly are acquired. Based on a virtual parameter fitting algorithm that dynamically correlates assembly and adjustment processes and parameters, the two types of data are fused with actual assembly and adjustment process operating condition information to perform standard range matching, interpolation completion, deviation calibration, and range constraint fitting, thus establishing a first multi-dimensional detection dataset. This includes: acquiring geometric and performance characteristic data through full-size 3D modeling and scanning of the drive motor assembly; collecting actual assembly and adjustment process operating condition information such as assembly tooling pressure, machining process speed, and feed rate; fusing the characteristic data with the operating condition information and completing the standard range interval matching mapping based on the virtual parameter fitting algorithm that dynamically correlates assembly and adjustment processes and parameters; and sequentially performing numerical interpolation completion, system deviation calibration, and range boundary constraint fitting processing on the fused data to form a structured first multi-dimensional detection dataset.

[0008] In one possible implementation, the first multi-dimensional detection dataset undergoes temporal identification of process interference features and invalid data masking to distinguish interference distortion and logical conflict data from different assembly and adjustment process stages. A second multi-dimensional detection dataset is obtained after dimensional alignment and format normalization. This includes: based on a multi-dimensional data interference feature clustering and outlier removal algorithm, the first multi-dimensional detection dataset is divided into dimensional features and clusters according to the assembly and adjustment process stages to distinguish between regular valid data and interference distortion and logical conflict data from the process stages; an interference judgment threshold and outlier verification rules related to the process stages are set, invalid data entries are marked and masked to obtain a relay multi-dimensional detection dataset; and the relay multi-dimensional detection dataset is fully aligned and standardized to generate the second multi-dimensional detection dataset.

[0009] In one possible implementation, a multi-parameter coupled judgment model is used to perform cross-validation correction on the process dimension, filter out abnormal parameters that do not match the assembly and adjustment process logic, and complete global deviation calibration to obtain a third multi-dimensional effective detection dataset. This includes: performing cross-validation on each parameter in the second multi-dimensional detection dataset item by item according to the process dimension, verifying the logical compatibility and numerical rationality between parameters in combination with the assembly and adjustment process logic, and determining the effective parameters; filtering out abnormal parameters that do not match the assembly and adjustment process logic, have logical contradictions, or whose values ​​exceed the standard range, performing full-dimensional global deviation calibration on the effective parameters, and unifying the numerical benchmark and dimension format to obtain the third multi-dimensional effective detection dataset.

[0010] In one possible implementation, a two-dimensional feature matching model is constructed based on a third-dimensional multi-dimensional effective detection dataset. Coupling sensitivity coefficients for geometric and performance parameters are introduced, and the dataset is substituted into the model for coupling analysis to calculate the coupling accuracy value. This includes: constructing a two-dimensional feature matching model containing an assembly accuracy feature analysis branch, an operational performance feature analysis branch, and a coupling accuracy quantification layer based on the parameter dimensions and feature attributes of the third-dimensional multi-dimensional effective detection dataset; introducing coupling sensitivity coefficients for geometric and performance parameters into the coupling accuracy quantification layer and setting parameter coupling correlation quantification rules; importing the third-dimensional multi-dimensional effective detection dataset into the model for multi-dimensional parameter feature matching and correlation coupling analysis; combining the coupling sensitivity coefficients to quantify the degree of collaborative adaptation between parameters; and outputting the coupling accuracy value.

[0011] In one possible implementation, the dataset is substituted for coupling analysis and the coupling accuracy value is calculated. The result is compared with a preset threshold. If the threshold is met, a basic test report is generated. If the threshold is not met, the effective detection data of the deviation and the coupling influence weight of the deviation parameters are output. This includes: comparing the calculated coupling accuracy value with the preset qualified threshold. If the threshold requirement is met, a basic test report containing the core detection parameters is generated. If the threshold requirement is not met, the deviation parameters are extracted and located from the third multi-dimensional effective detection dataset, the effective detection data of the deviation is output, and the coupling influence weight of each deviation parameter is calculated and output based on the coupling sensitivity coefficient.

[0012] In one possible implementation, based on deviation data, deviation amplitude, coupling influence weights, and preset process traceability rules, the causes and calibration parameters of the core assembly and adjustment processes are reverse-engineered. The motor assembly undergoes simulation-based iterative correction of the assembly and adjustment processes. Data with remaining deviations is then compensated for using coupling techniques to obtain a fourth multi-dimensional detection dataset. This includes: decomposing valid deviation detection data into single-dimensional independent parameter items; matching parameters with the correlation characteristics of the assembly and adjustment processes based on deviation amplitude and coupling influence weights; and reverse-engineering the nodes and causes of the core assembly and adjustment processes based on preset process traceability rules. Combined with preset judgment rules and process characteristics, standardized process calibration parameters corresponding to the causes of the core assembly and adjustment processes are calculated. Based on the calibration parameters, the motor assembly undergoes simulation-based iterative correction of the assembly and adjustment processes, simulating multiple rounds of process parameter adjustments and verifying the effects to determine valid detection data with remaining deviations. According to the coupling influence weights, the associated geometric and performance characteristic data are adjusted synchronously, and coupling compensation is applied to the remaining deviation data to calibrate the parameter values ​​to the standard range. The detection data with completed deviation compensation is then merged and consolidated with the original qualified detection data to obtain the fourth multi-dimensional detection dataset.

[0013] In one possible implementation, the fourth multi-dimensional detection dataset is used to back-substitute the model to recalculate the coupling accuracy value. The effectiveness of the correction is verified based on the change in accuracy value between the two calculations. A detection report is generated by integrating the entire process of detection data, causes of deviation, corrective measures, and verification results, including: The fourth multi-dimensional detection dataset is substituted back into the two-dimensional feature matching model, and the coupling accuracy value is recalculated through the repeatability accuracy calculation process. The change in the coupling accuracy value between the two calculations is calculated and compared with the preset correction effectiveness threshold to verify the actual effectiveness of simulation iterative correction and coupling compensation correction. The entire process of detection data, deviation causes, simulation iterative correction steps, coupling compensation correction parameters, and correction effectiveness verification results are integrated to generate a detection report containing full-dimensional detection information in a standardized format. Secondly, embodiments of this application provide a drive motor assembly detection device, which includes a model extraction section, a parameter processing section, and a detection result generation section.

[0014] Secondly, embodiments of this application provide a drive motor assembly testing device, including: a model extraction part, a parameter processing part, and a test result generation part.

[0015] Thirdly, embodiments of this application provide an electronic device, which includes: The memory, processor, and computer program stored in the memory, wherein the processor is configured to run the computer program to execute the drive motor assembly detection method.

[0016] In the above setup, multi-dimensional core parameters are collected across the entire domain through 3D modeling to construct a complete original detection dataset, avoiding errors from physical hardware contact measurements. After process interference feature removal and dual-layer calibration of the multi-parameter coupled judgment model, a clean and valid dataset is obtained, eliminating data distortion interference with subsequent detection from the source. A dual-dimensional feature matching model transforms scattered parameters into quantified coupled precision values, achieving objective and standardized judgment, replacing subjective human experience, and accurately locking down deviation data. For non-conforming samples, deviation process tracing, virtual assembly and adjustment correction, and dynamic compensation are completed, filling the gap in traditional detection without corresponding correction paths, achieving a complete closed loop from detection to correction. Finally, the model is reused for retesting and a fully traceable report is generated. No physical testing equipment is required throughout the process, significantly reducing detection costs. It balances detection accuracy, process controllability, and result traceability, effectively solving the pain points of large errors, subjective judgment, and fragmented processes in traditional detection. It is suitable for teaching motor assemblies in multiple scenarios, thereby improving the accuracy of determining the overall suitability of the motor assembly. Attached Figure Description

[0017] Figure 1 This is a structural block diagram of an electronic device according to an embodiment of this application.

[0018] Figure 2 This is a flowchart of a drive motor assembly testing method according to an embodiment of this application.

[0019] Figure 3 This is a structural block diagram of a drive motor assembly testing device according to an embodiment of this application. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present application, the technical solutions in specific embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in specific embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0022] It should be noted that the terms "first," "second," and similar terms used in this application specification and claims do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, "an" or "a" and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. "A plurality of" indicates at least two. "Comprising" and similar terms mean that the elements or objects preceding "comprising" cover the elements or objects listed after "comprising" and their equivalents, and do not exclude other elements or objects. "Connection" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0023] The singular forms “a,” “the,” and “the” used in this application specification and appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0024] The train traction power supply control method provided in this embodiment can be executed in electronic device 100 or similar device. Figure 1 This is a hardware structure block diagram of an electronic device 100 that implements an embodiment of this application. For example... Figure 1 As shown, the electronic device 100 may include one or more ( Figure 1 (Only one is shown) Memory 12 and processor 11. This electronic device 100 is the control terminal of the train system, which is used to detect the parameters of the drive motor assembly, thereby determining whether the motor assembly as a whole is compatible.

[0025] The memory 12 stores program instructions, such as application software programs and modules, like the computer program for a train traction power supply control method in this embodiment. The processor 11 executes the program instructions stored in the memory 12. By running the computer program stored in the memory 12, it can perform various functional applications and data processing, such as acquiring different types of parameters of the drive motor assembly.

[0026] The processor 11 may include, but is not limited to, a microprocessor 11 (Microcontroller Unit, abbreviated as MCU) or a programmable gate array (FPGA).

[0027] Those skilled in the art will understand that Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the electronic device 100 described above. For example, the electronic device 100 may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown are illustrated.

[0028] In some embodiments, a portion of the following method for detecting the drive motor assembly of a train is performed in electronic device 100, while the remainder of the method is performed in parameter processing section to achieve drive motor assembly detection.

[0029] This embodiment also provides a method for testing a drive motor assembly, such as... Figure 2 As shown, Figure 2 This is a flowchart of a drive motor assembly testing method according to an embodiment of this application. This method aims to improve the accuracy of motor assembly testing. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention. The specific steps of this drive motor assembly testing method are as follows: S1 acquires the geometric and performance characteristic data of the drive motor assembly. Based on the virtual parameter fitting algorithm that dynamically associates the assembly and adjustment process and parameters, it integrates the actual assembly and adjustment process working condition information to perform standard range matching, interpolation completion, deviation calibration and range constraint fitting on the geometric and performance characteristic data, and establishes the first multi-dimensional detection dataset.

[0030] In the above steps, multi-dimensional detection parameters refer to various quantitative indicators that can comprehensively characterize the core state of the new energy vehicle drive motor assembly, not single-dimensional parameters. In this application, multi-dimensional detection parameters include parameters of electrical insulation performance, sealing and airtightness performance, shaft mechanical geometric tolerances, and the three-dimensional morphology of core components. That is, multi-dimensional detection parameters can directly reflect the assembly accuracy and basic performance of the motor assembly. The first multi-dimensional detection dataset is the initial raw dataset of this detection process. It is a complete data set formed by organizing and collecting the scattered multi-dimensional detection parameters according to a unified data format specification and parameter classification logic. This dataset has not undergone any screening, purification, correction, or calculation processing, and completely retains the initial state of the parameters extracted from the three-dimensional model scanning and modeling, so as to improve the comprehensiveness of the source of drive motor assembly detection parameters, thereby facilitating subsequent detection of the drive motor assembly.

[0031] In this embodiment, multi-dimensional detection parameters are extracted and obtained through scanning and modeling a 3D model of the drive motor assembly. This eliminates the need for physical insulation testers, airtightness testers, geometric tolerance measuring instruments, and other hardware. Parameter acquisition is completed entirely using virtual modeling data, fundamentally avoiding tooling interference and measurement errors caused by physical hardware acquisition. By establishing the original data carrier for the detection process, complete data support is provided for all subsequent data processing, accuracy calculation, and deviation determination steps, facilitating subsequent testing of the drive motor assembly.

[0032] As one implementation method, the specific steps in step S1 to obtain the geometric and performance characteristic data of the drive motor assembly, and to establish the first multi-dimensional detection dataset by integrating the actual assembly and adjustment process and parameter dynamic correlation virtual parameter fitting algorithm, and performing standard range matching, interpolation completion, deviation calibration, and range constraint fitting on the two types of data, include: S11 acquires geometric and performance feature data through full-size 3D modeling and scanning of the drive motor assembly, and collects actual assembly and adjustment process condition information such as assembly tooling pressure, machining speed and feed rate.

[0033] S12 is a virtual parameter fitting algorithm based on the dynamic correlation between assembly and adjustment processes and parameters. It integrates feature data with working condition information and completes the standard range interval matching mapping.

[0034] S13 sequentially performs numerical interpolation, system bias calibration, and range boundary constraint fitting on the fused data to form a structured first multi-dimensional detection dataset. Geometric and performance characteristic data are obtained through a 3D model of the drive motor assembly. Geometric characteristic data includes shaft dimensional tolerances, relative component positions, and 3D topographic contour data. Performance characteristic data includes electrical insulation resistance, airtightness leakage rate, and mechanical stiffness parameters. After data acquisition, a virtual parameter fitting algorithm is initiated. The algorithm's construction process involves building a data preprocessing module, a standard interval mapping module, and a numerical optimization module, and presetting the standard range intervals, allowable system deviation ranges, and range boundary constraints for each parameter of the motor assembly.

[0035] Based on the constructed virtual parameter fitting algorithm, geometric feature data and performance feature data are imported into the standard interval mapping module of the algorithm to complete the matching mapping between parameters and standard range intervals. Then, the numerical optimization module performs three fine-tuning processes on the two types of data: numerical interpolation completion (for missing scattered parameters), system deviation calibration (for systematic small errors generated by scanning modeling), and range boundary constraint fitting (to prevent parameters from exceeding the standard range). After processing, the data is regularized to form a standardized first multi-dimensional detection dataset.

[0036] In the above setup, the normalization process of the virtual parameter fitting algorithm can optimize the regularity and completeness of the initial dataset, avoid excessive invalid data due to missing parameters or exceeding the range in subsequent steps, reduce the workload of data processing, and improve the computational efficiency of subsequent multi-parameter coupling judgment model and two-dimensional feature matching model, thus achieving synergistic efficiency between the initial data stage and subsequent processing stages.

[0037] S2 performs time-series identification of process interference features and invalid data masking on the first multi-dimensional detection dataset, distinguishes interference distortion and logical conflict data in different assembly and adjustment process stages, obtains the second multi-dimensional detection dataset after dimension alignment and format normalization, and then performs process dimension cross-validation correction on it through a multi-parameter coupled judgment model, filters out abnormal parameters that do not match the logic of the assembly and adjustment process and completes global deviation calibration to obtain the third multi-dimensional effective detection dataset.

[0038] The above steps include process interference feature identification and invalid data masking. Process interference features refer to abnormal parameter characteristics caused by assembly defects and performance failures not inherent to the drive motor itself. These are mostly numerical distortions, logical conflicts, and format errors generated during virtual modeling and data collection, and do not belong to quality problems of the motor itself. Invalid data refers to abnormal parameter data that matches the above process interference features and has no detection reference value, and does not represent the actual performance and assembly status of the motor. By screening all parameters in the first multi-dimensional detection dataset for process interference features, invalid data can be selectively masked to retain valid parameters with real reference value, thereby obtaining the second multi-dimensional detection dataset.

[0039] In this application, a multi-parameter coupling judgment model is used for cross-validation correction. This model is an analytical tool used to verify the logical correlation and numerical rationality between multi-dimensional parameters, thus identifying contradictions and independent anomalies among them. Using the second multi-dimensional detection dataset as the processing object, the multi-parameter coupling judgment model is used to cross-verify each parameter within the dataset, checking the compatibility and compliance between different dimensions of parameters. This further eliminates remaining abnormal data that has not been initially masked, while simultaneously standardizing the numerical benchmarks and format specifications of various parameters. After global calibration, a calibrated third multi-dimensional effective detection dataset is obtained. This dataset is free of interference information, abnormal data, and logical contradictions; the parameters are authentic and reliable, and the format is uniform and standardized, making it directly usable for subsequent coupling accuracy value calculations.

[0040] The above steps, through a two-layer progressive data processing logic, comprehensively improve the quality of the original data, solving the problems of the first multi-dimensional detection dataset being messy, distorted, and unusable. At the same time, they lay a solid data foundation for the subsequent calculation of coupling accuracy values, avoiding situations where missing processes interfere with feature recognition, invalid data masking, and cross-validation correction, resulting in deviations in the subsequent calculation of coupling accuracy values ​​due to invalid data interference, thus failing to achieve accurate detection.

[0041] As one implementation method, the basic construction process of the multi-parameter coupling judgment model is as follows: a three-layer basic framework of the model is built, namely the parameter input layer, the coupling verification layer, and the data output layer; the parameter logical matching rules, numerical standard interval thresholds, and dimensional correlation verification algorithms are built into the coupling verification layer, the allowable range of numerical deviations of the same type of parameters and the correlation constraints between parameters of different dimensions are set to ensure that the model can automatically identify contradictory and abnormal items between parameters.

[0042] Based on the completed multi-parameter coupling judgment model, the specific implementation process of this step is as follows: First, the first multi-dimensional detection dataset is fully traversed to identify process interference features one by one, mark and mask all invalid data entries, and perform preliminary reorganization of the valid parameters after removing interference to obtain the second multi-dimensional detection dataset; then, the second multi-dimensional detection dataset is imported into the parameter input layer of the multi-parameter coupling judgment model, and the parameters are cross-verified item by item through the coupling verification layer to filter out hidden abnormal parameters. Then, the remaining valid parameters are calibrated in terms of numerical benchmark and dimension format. Finally, the data output layer outputs the calibrated third multi-dimensional valid detection dataset. This dataset is free of interference, abnormalities, and logical contradictions, and can be directly used for subsequent coupling accuracy calculation. The specific steps for obtaining the second multi-dimensional detection dataset in step S2 include: S21 is based on a multi-dimensional data interference feature clustering and outlier screening algorithm. It performs dimensional feature extraction and clustering on the first multi-dimensional detection dataset according to the assembly and adjustment process stage, distinguishing between regular valid data and interference distortion and logical conflict data in the process stage.

[0043] S22 sets the interference judgment threshold and outlier verification rules associated with the process, marks and blocks invalid data entries, and obtains the relay multi-dimensional detection dataset.

[0044] S23 performs full-dimensional alignment and standardization on the relay multi-dimensional detection dataset to generate a second multi-dimensional detection dataset.

[0045] In the above settings, based on the first multi-dimensional detection dataset, a multi-dimensional data interference feature clustering and outlier screening algorithm is enabled. The construction process of this algorithm is as follows: build an algorithm feature extraction module, a clustering module, a threshold determination module, and a data straightening module; the feature extraction module has built-in parameter dimension splitting rules, the clustering module has built-in K-means clustering algorithm, the threshold determination module has preset interference determination thresholds and outlier verification rules, and the data straightening module has built-in dimension alignment and format unification rules.

[0046] Based on the completed algorithm, the specific implementation process of this step is as follows: First, the feature extraction module extracts dimensional features from various parameters in the first multi-dimensional detection dataset. Then, the clustering module divides the parameters into three categories: regular valid data, interference and distortion data in the virtual detection scenario, and logical conflict data. Subsequently, the threshold judgment module compares various parameters with the interference judgment threshold, marks and filters out invalid data entries that exceed the threshold range, and obtains the relay multi-dimensional detection dataset after the filtering is completed. Finally, the relay multi-dimensional detection dataset is imported into the data straightening module to complete the parameter dimension alignment and format straightening, generating a standardized second multi-dimensional detection dataset.

[0047] In the above settings, this step uses the clustering and threshold determination functions of a dedicated interference removal algorithm to accurately remove invalid data, resulting in higher interference identification accuracy. It can completely remove explicit abnormal data, reducing the computational burden on the subsequent cross-validation correction of the multi-parameter coupled determination model. At the same time, it significantly improves the purity of the second multi-dimensional detection dataset, further ensuring the quality of the subsequent third multi-dimensional effective detection dataset.

[0048] As one implementation method, step S2 involves performing cross-validation correction on the process dimension using a multi-parameter coupled judgment model, filtering out abnormal parameters that do not match the assembly and adjustment process logic, and completing global deviation calibration to obtain the third multi-dimensional effective detection dataset. S2a performs cross-verification of each parameter in the second multi-dimensional detection dataset along the process dimension, and combines the assembly and adjustment process logic to verify the logical compatibility and numerical rationality between parameters, thereby determining the effective parameters.

[0049] S2b filters out abnormal parameters that do not match the assembly and adjustment process logic, have logical contradictions, or have values ​​that exceed the standard range. It performs full-dimensional global deviation calibration on the effective parameters, unifies the numerical benchmark and dimensional format, and obtains a third multi-dimensional effective detection dataset.

[0050] As one implementation approach, the above steps can further refine the construction details of the multi-parameter coupled judgment model by adding a parameter classification verification submodule and a global deviation calibration submodule to the original coupling verification layer, thus clarifying the cross-verification logic of the model. After importing the second multi-dimensional detection dataset into the model, the parameter classification verification submodule first performs cross-verification on each parameter, checking the numerical consistency of parameters of the same type and the logical adaptability of parameters of different dimensions, determining the valid parameters, and directly filtering out implicit abnormal parameters with logical contradictions or values ​​exceeding the standard range; then, the global deviation calibration submodule performs global calibration on the remaining valid parameters, unifying the numerical benchmark, unit of measurement, and dimensional format of parameters of the same type, and eliminating scattered deviations between parameters; after calibration, a third multi-dimensional valid detection dataset without any abnormalities and with a completely unified format is output.

[0051] S3 constructs a two-dimensional feature matching model based on the third multi-dimensional effective detection dataset. It introduces the coupling sensitivity coefficient of geometric and performance parameters, substitutes them into the dataset for coupling analysis and calculates the coupling accuracy value. The coupling accuracy value is compared with a preset threshold. If the threshold is met, a basic detection report is generated. If the threshold is not met, deviation data is extracted from the third multi-dimensional effective detection dataset, the deviation magnitude of each deviation data is calculated, and the coupling influence weight corresponding to each deviation data is calculated and output based on the coupling sensitivity coefficient.

[0052] The coupling accuracy value is a core quantitative indicator output by the two-dimensional feature matching model. Its magnitude directly reflects the degree of coordination and adaptation between various components and parameters of the drive motor assembly, and is the core basis for determining whether the motor assembly is qualified. In the above steps, the two-dimensional feature matching model is a quantitative analysis model built specifically for the drive motor assembly. Through the core performance and assembly feature dimensions of the drive motor assembly, it can transform the scattered parameters of the third-dimensional effective detection dataset into a single quantifiable judgment indicator, realizing the transformation from abstract parameters to intuitive results.

[0053] It should be clarified that the valid detection data with deviations belong to the valid parameter units in the third multi-dimensional valid detection dataset. This type of data truly reflects the actual state of the motor assembly. Only the value does not reach the preset standard range. It can accurately point to the specific deviation problem at the level of motor assembly or performance. It is the core target basis for subsequent core assembly and adjustment processes to trace the cause and dynamically compensate and correct.

[0054] In this embodiment, this step first completes the standardized construction of the two-dimensional feature matching model, then fully substitutes the calibrated third-dimensional effective detection dataset into the two-dimensional feature matching model, and obtains the coupling accuracy value through model calculation. Subsequently, the coupling accuracy value is precisely compared with a preset threshold, forming two parallel branch logics: if the coupling accuracy value reaches the preset threshold, a basic detection report is directly generated, completing the basic detection process; if the coupling accuracy does not reach the preset threshold, report generation is paused, and effective detection data with deviations in the drive motor assembly are accurately extracted and output. This step transforms manual experience judgment into standardized model calculation, which can improve the objectivity and consistency of the detection results, and at the same time realize branching between qualified and unqualified scenarios, which is conducive to subsequent closed-loop correction of effective detection parameters.

[0055] As one implementation method, a dual-branch core framework is constructed, consisting of an assembly accuracy feature analysis branch and an operational performance feature analysis branch, with the two branches connected to a coupling accuracy quantification layer. The assembly accuracy feature analysis branch incorporates geometric tolerance and component shape matching algorithms, while the operational performance feature analysis branch incorporates electrical insulation and airtightness calculation algorithms. The coupling accuracy quantification layer pre-sets weight allocation rules and a comprehensive calculation formula, weighting and fusing the analysis results from both branches to output a coupling accuracy value in the range of 0-100%. A higher value indicates better compatibility and coordination among the components and parameters of the motor assembly.

[0056] Based on the completed two-dimensional feature matching model, the specific implementation process of this step is as follows: The third multi-dimensional effective detection dataset is synchronously imported into the two major analysis branches of the model to complete the feature extraction and analysis of assembly accuracy and operational performance, respectively. Then, the coupling accuracy value is calculated through the coupling accuracy quantization layer. The calculation result is precisely compared with a preset threshold. If the coupling accuracy value is greater than or equal to the preset threshold, the motor assembly is deemed qualified, and a basic inspection report is directly generated. If the coupling accuracy value is less than the preset threshold, the motor assembly is deemed unqualified. The effective detection data with deviations is accurately located and output from the third multi-dimensional effective detection dataset. This type of data is a true reflection of the effective parameters of the motor assembly defects, rather than interference data, and can be directly used as the target basis for subsequent deviation correction. Step S3, which involves constructing a two-dimensional feature matching model based on the third multi-dimensional effective detection dataset, introducing the coupling sensitivity coefficients of geometric and performance parameters, substituting them into the dataset for coupling analysis, and calculating the coupling accuracy value, includes the following steps: S31 constructs a two-dimensional feature matching model based on the parameter dimensions and feature attributes of the third multi-dimensional effective detection dataset, which includes an assembly accuracy feature analysis branch, an operational performance feature analysis branch, and a coupled accuracy quantization layer.

[0057] S32 introduces coupling sensitivity coefficients for geometric and performance parameters in the coupling precision quantization layer, and sets the parameter coupling correlation quantization accounting rules.

[0058] S33 imports the third multi-dimensional effective detection dataset into the model to perform multi-dimensional parameter feature matching and correlation coupling analysis. It combines the coupling sensitivity coefficient to quantify the degree of synergistic adaptation between parameters and outputs the coupling accuracy value.

[0059] In this implementation, based on the parameter dimensions and feature attributes of the third multi-dimensional effective detection dataset, a refined two-dimensional feature matching model is constructed: First, the model's framework structure is determined according to the parameter types of the dataset, and the internal algorithms of the assembly accuracy feature branch and the operating performance feature branch are refined. For the four core parameters of geometric tolerance, three-dimensional morphology, insulation performance, and airtightness performance, exclusive feature extraction rules are configured respectively. Then, model feature adaptation settings are performed. According to the motor assembly detection requirements, the weight ratio of the two feature branches is adjusted, and the calculation formula of the coupling accuracy quantization layer is optimized to ensure that the model is fully adapted to the current dataset.

[0060] After the model is built, the third multi-dimensional effective detection dataset is completely imported into the model, and multi-dimensional parameter feature matching and correlation coupling analysis is started to extract the core feature values ​​of each parameter and analyze the collaborative correlation between parameters. Based on the feature data obtained from the analysis, the degree of collaborative adaptation between each parameter of the motor assembly is fully quantified and calculated through the coupling accuracy quantification layer. Finally, the coupling accuracy value is determined and output through the calculation results. This value can intuitively and accurately reflect the overall assembly adaptability of the motor assembly.

[0061] In the above settings, this step combines the actual attributes of the third multi-dimensional effective detection dataset to customize the model, achieving a high degree of data and model adaptation. The high-quality dataset output by the previous steps provides support for the accurate calculation of the model, and the refined construction of the model can maximize the mining of data value. The synergy between the two makes the calculation result of the coupling accuracy value more consistent with the actual state of the motor assembly, greatly improving the accuracy of detection and judgment.

[0062] As one implementation method, step S3, which involves substituting the dataset for coupling analysis and calculating the coupling accuracy value, comparing it with a preset threshold, generates a basic detection report if the threshold is met, and outputs the effective detection data of the deviation and the coupling influence weight of the deviation parameters if the threshold is not met. S3a compares the calculated coupling accuracy value with the preset qualified threshold. If the threshold requirement is met, a basic test report containing the core test parameters is generated.

[0063] If S3A does not meet the threshold requirement, it extracts and locates the deviation parameters from the third multi-dimensional effective detection dataset, outputs the effective detection data of the deviation, and calculates and outputs the coupling influence weight of each deviation parameter based on the coupling sensitivity coefficient.

[0064] In this embodiment, the coupling accuracy value calculated by the two-dimensional feature matching model is compared one-to-one with a preset qualified threshold. The preset qualified threshold is pre-entered into the detection system and remains fixed throughout the process without temporary adjustments. If the coupling accuracy value is greater than or equal to the preset threshold, the motor assembly is directly determined to be qualified, and the system report template is retrieved to generate a test report containing basic test data and a qualified conclusion. If the coupling accuracy value is less than the preset threshold, the motor assembly is determined to be unqualified, and the deviation parameter positioning module is activated. From the third multi-dimensional effective test data set, parameters whose values ​​do not meet the standard and affect the coupling accuracy are extracted one by one. The dimension and corresponding position of each deviation parameter are accurately located, and such parameters are output as valid test data for deviations in the drive motor assembly.

[0065] In the above setup, this step relies on the precise coupling accuracy value calculated in the previous step to achieve rapid separation of qualified and unqualified samples. Qualified samples can directly complete the testing process, while the deviation data of unqualified samples are accurately located and highly targeted, requiring no additional screening. This directly provides a clear basis for deviation tracing and correction in the subsequent S4, effectively shortening the processing cycle of unqualified samples and improving the overall testing efficiency.

[0066] Based on deviation data, deviation magnitude, coupling influence weight, and preset process traceability rules, S4 reverse-engineers the causes and calibration parameters of the core assembly and adjustment process, performs simulation iteration correction of the motor assembly, performs coupling compensation correction on the data that still has deviations, and obtains the fourth multi-dimensional detection dataset.

[0067] The above steps are only executed for unqualified scenarios where the coupling accuracy value in step S3 does not reach the preset threshold. For qualified scenarios, this step is not required to achieve a closed-loop process of deviation location, source tracing, virtual correction, and data compensation. The core assembly and adjustment process cause refers to the fundamental assembly and adjustment process problem that leads to the existence of deviations in valid detection data and insufficient coupling accuracy. It is not a superficial numerical deviation, but rather a source process defect causing the deviation. The process calibration parameters are standardized adjustment parameters formulated according to industry standards and assembly specifications, targeting the core assembly and adjustment process cause, and are used to guide the assembly and adjustment correction work of the drive motor assembly.

[0068] In this embodiment, this step first relies on the valid detection data with deviations output in step S3 to match the corresponding correlation between deviation characteristics and assembly / adjustment processes. It traces the root cause of each valid detection data point with deviations to the core assembly / adjustment process, avoiding blind investigation. Then, based on the traced root causes of the core assembly / adjustment processes, it calculates the corresponding standardized process calibration parameters and performs assembly / adjustment correction on the drive motor assembly according to these parameters. After the assembly / adjustment correction is completed, it again identifies the corresponding valid detection data with deviations and performs dynamic compensation correction on this type of data, calibrating the parameter values ​​to the preset standard acceptable range to eliminate data deviations. Finally, it integrates and aggregates the qualified parameters that have undergone dynamic compensation correction with the original qualified parameters that did not show deviations, unifying the data format and numerical benchmark to form the fourth multi-dimensional detection dataset after deviation data compensation. All parameters in this dataset meet the standard requirements, and there are no valid detection data points with deviations; it is a corrected and complete compliant dataset.

[0069] By implementing the above settings, the technical gap of simply judging non-compliance without making corrections, which is common in traditional testing, is avoided. By combining the root cause tracing and dynamic compensation correction of core assembly and adjustment processes, valid test data with deviations are optimized into qualified parameters, providing a compliant data basis for subsequent secondary coupling accuracy value calculation. At the same time, the source of deviation is identified to prevent the recurrence of similar deviations.

[0070] As one implementation method, step S4, based on deviation data, deviation magnitude, coupling influence weight, and preset process traceability rules, reverse-engineers the causes and calibration parameters of the core assembly and adjustment processes, performs simulation iteration correction of the motor assembly assembly, and performs coupling compensation correction on the data that still has deviations, to obtain the fourth multi-dimensional detection dataset includes: S41 decomposes the effective deviation detection data into single-dimensional independent parameter items, combines the deviation amplitude, coupling influence weight matching parameters and the correlation characteristics of the assembly and adjustment process, and reverse deduces the core assembly and adjustment process nodes and causes based on the preset process traceability rules.

[0071] S42 combines preset judgment rules and process characteristics to calculate standardized process calibration parameters for the causes of corresponding core assembly and adjustment processes.

[0072] S43 performs simulation and iterative correction of the motor assembly assembly process based on calibration parameters, simulates multiple rounds of process parameter adjustments and verifies the effect, and determines the valid test data that still has deviations.

[0073] S44 synchronously adjusts the associated geometric and performance characteristic data according to the coupling influence weight, performs coupling compensation correction on the still biased data, and calibrates the parameter values ​​to the standard range; the test data with completed deviation compensation is merged and integrated with the original qualified test data to obtain the fourth multi-dimensional test dataset.

[0074] In this implementation, the output valid detection data with deviations is decomposed into independent single-dimensional parameter items to avoid traceability errors caused by multiple mixed parameters. A database linking deviation parameters and processes is built, containing built-in information on the corresponding association between standard parameter ranges, deviation characteristics, and assembly / adjustment processes. The decomposed independent single-dimensional parameter items are matched with the database content to locate the parameter deviation magnitude and the corresponding process association characteristics, thus identifying the specific process node that caused the deviation in the valid detection data.

[0075] Combining the preset assembly judgment rules with the causes of core assembly and adjustment processes under the corresponding process nodes, the system automatically calculates the standardized process calibration parameters for the causes of the corresponding core assembly and adjustment processes. The calibration parameters are consistent with the actual assembly specifications and can be directly used for virtual assembly and adjustment correction. Based on the standardized process calibration parameters, the drive motor assembly is virtually assembled and adjusted to simulate actual assembly and debugging operations and repair process defects. After the virtual assembly and adjustment is completed, the parameters are verified again to determine the valid detection data that still has deviations. Dynamic error compensation correction is performed on this type of data, and the parameters are calibrated to the standard range through numerical fine-tuning.

[0076] Finally, all the test data that have undergone deviation compensation are integrated and consolidated with the original qualified test data that did not show deviations. The data format, numerical benchmarks and dimensional information are unified to obtain a complete and standardized fourth multi-dimensional test dataset. This dataset is a fully qualified dataset and can be directly used for subsequent secondary accuracy retesting.

[0077] In the above setup, this step takes the output deviation data as the starting point and the correction logic as the core. Through four operations—parameter decomposition, process traceability, virtual assembly adjustment, and dynamic compensation—the root cause of the deviation is repaired. The deviation data locked in the previous steps provides a target for correction, and the correction results in this step provide qualified data for subsequent secondary retesting. The preceding and following steps form a complete "deviation, correction, and compliance" logical chain. At the same time, the assembly process is optimized at the process level, achieving two-way collaboration between inspection and assembly optimization.

[0078] S5 uses the fourth multi-dimensional detection dataset to back-substitute the model to recalculate the coupling accuracy value. Based on the change in accuracy value between the two calculations, it verifies the effectiveness of the correction. It integrates the full-process detection data, the causes of deviation, the correction measures, and the verification results to generate a detection report, thereby realizing the detection of the drive motor assembly.

[0079] The above steps utilize the fourth multi-dimensional detection dataset after deviation data compensation, and continue using the two-dimensional feature matching model without requiring model reconstruction, ensuring consistency in the detection standards. This step, after substituting the fourth multi-dimensional detection dataset after deviation data compensation into the two-dimensional feature matching model, repeats the coupling accuracy value calculation process, calculating the coupling accuracy value of the motor assembly after assembly adjustment correction and dynamic compensation correction, verifying the actual effect of the correction operation, and confirming whether the motor assembly meets the standards.

[0080] After the coupling accuracy value is calculated, the entire process of testing information is integrated to generate a test report containing test data, causes of deviations, and corrective measures. The report covers initial multi-dimensional test parameters, details of the first multi-dimensional test dataset, process interference screening results, second multi-dimensional test dataset, third multi-dimensional valid test dataset, initial calculation results of coupling accuracy value, details of valid test data with deviations, causes of core assembly and adjustment processes, process calibration parameters, assembly and adjustment correction measures, dynamic compensation correction content, fourth multi-dimensional test dataset after deviation data compensation, and secondary coupling accuracy value calculation results. The information is comprehensive, complete, traceable, and archiveable. Regardless of whether there are initial deviations in the motor assembly, a complete and standardized test report can be generated, fully completing the virtual testing of the entire process of the new energy vehicle drive motor assembly.

[0081] In this application, by calculating the accuracy value of the secondary coupling, it is confirmed that the performance of the motor assembly after the adjustment correction and dynamic compensation correction has reached the preset state. At the same time, the entire testing process can be closed, thereby improving the accuracy and reliability of the motor assembly testing results.

[0082] As one implementation method, step S5 involves recalculating the coupling accuracy value by substituting the fourth multi-dimensional detection dataset back into the model, verifying the effectiveness of the correction based on the change in accuracy value between the two calculations, and integrating the entire process detection data, causes of deviation, correction measures, and verification results to generate a detection report. S51 substitutes the fourth multi-dimensional detection dataset back into the two-dimensional feature matching model, and the repeatability accuracy calculation process recalculates the coupling accuracy value; it calculates the change in the coupling accuracy value between the two calculations, compares it with the preset correction effectiveness threshold, and verifies the actual effectiveness of simulation iterative correction and coupling compensation correction; it integrates the full-process detection data, deviation causes, simulation iterative correction steps, coupling compensation correction parameters, and correction effectiveness verification results, and generates a detection report containing full-dimensional detection information in a standardized format.

[0083] In this embodiment, the two-dimensional feature matching model that has been built and debugged is reused without rebuilding the framework or adjusting the parameters, ensuring that the standards of the two accuracy calculations are completely consistent. The fourth multi-dimensional detection dataset that has been completed in the previous deviation compensation is substituted back into the model to completely repeat the coupling accuracy calculation process and calculate the coupling accuracy value of the motor assembly again. The calculation results are used to verify the actual effect of the assembly adjustment correction and dynamic compensation, ensuring that the corrected motor assembly fully meets the standards.

[0084] After the secondary accuracy calculation is completed, the report integration module is activated to comprehensively collect all the testing data from the entire process, including the initial multi-dimensional testing parameters of S1, the first multi-dimensional testing dataset, the second multi-dimensional testing dataset, the third multi-dimensional testing dataset, the initial coupling accuracy value, deviation parameters, the causes of core assembly and adjustment processes, process calibration parameters, assembly and adjustment correction measures, dynamic compensation content, and the secondary coupling accuracy value. The above information is integrated according to the standard template to generate a standardized testing report containing the complete testing process, the causes of deviations, and correction measures, so as to help determine the influencing factors of the drive motor assembly during the testing process.

[0085] In the above setup, this step reuses the constructed two-dimensional feature matching model and relies on the fourth multi-dimensional detection dataset that has been corrected in the previous step to achieve secondary accuracy retesting and report generation. This not only verifies the effectiveness of the previous correction step, but also completes the closed-loop conclusion of the entire detection process. All steps in the entire process are connected and support each other, and model construction and data processing work together to ultimately achieve the core effects of standardized detection process, accurate results, and closed-loop traceability.

[0086] In summary, the drive motor assembly testing method in this application completes the acquisition of multi-dimensional testing parameters and constructs the first multi-dimensional testing dataset in step S1. This provides the sole original data source for subsequent full-process data processing, accuracy calculation, and deviation analysis, and directly lays the foundation for the start of the testing process as there is no basis for subsequent data processing and testing judgment steps. Secondly, using the first multi-dimensional testing dataset output in step S1 as the processing object, step S2 sequentially completes the identification of process interference features and the masking of invalid data. Then, through cross-validation and correction by the multi-parameter coupling judgment model, noise information and hidden abnormal parameters in the original data are eliminated step by step, finally obtaining the calibrated third multi-dimensional valid testing dataset. This eliminates parameter interference caused by defects not inherent to the motor itself, thereby improving the numerical accuracy of subsequent coupling accuracy calculation and avoiding the distortion of drive motor assembly testing conclusions due to invalid data.

[0087] Furthermore, based on the high-quality third-dimensional effective detection dataset output in step S2, a two-dimensional feature matching model is constructed in step S3 to calculate the coupling accuracy value. This transforms scattered multi-dimensional parameters into a single quantifiable judgment index, which, combined with a preset threshold, completes the qualification judgment. This achieves standardized and objective judgment of the detection results, replacing subjective experience judgment, and accurately identifies and outputs effective detection data with deviations, thereby improving the accuracy of acquiring multi-dimensional effective detection data. For the data with deviations output in step S3, step S4 traces the cause of the corresponding core assembly and adjustment process, completes the assembly and adjustment correction based on the process calibration parameters, and then implements dynamic compensation for the deviation data, ultimately forming a fourth-dimensional detection dataset after deviation data compensation. This step achieves precise correction through the deviation target determined in the previous step, realizing the core effect of "deviation location-closed-loop correction".

[0088] Furthermore, based on the fourth multi-dimensional detection dataset corrected in step S4, the two-dimensional feature matching model is reused in step S5 to complete the secondary coupling accuracy calculation, verify the actual effectiveness of the correction operation, and then generate a complete detection report containing full-process information. This completes the final detection loop by connecting the previous correction results, ensuring the rigor of the final detection conclusion and forming an archiveable and traceable detection document, thereby further improving the accuracy of motor assembly detection.

[0089] like Figure 3 As shown, this application also includes a drive motor assembly testing device 200, which includes a model extraction part 21, a parameter processing part 22, and a testing result generation part 23.

[0090] In this application, the drive motor assembly testing device may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This drive motor assembly testing device is merely an example and should not impose any limitations on the functionality and scope of the embodiments of the present invention.

[0091] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for testing a drive motor assembly for new energy vehicles, characterized in that, The method includes: Obtain geometric and performance characteristic data of the drive motor assembly; based on the virtual parameter fitting algorithm that dynamically associates the assembly and adjustment process and parameters, integrate the actual assembly and adjustment process working condition information to perform standard range matching, interpolation completion, deviation calibration and range constraint fitting on the geometric and performance characteristic data to establish a first multi-dimensional detection dataset. The first multi-dimensional detection dataset is subjected to temporal identification of process interference features and invalid data masking. Interference distortion and logical conflict data in different assembly and adjustment process stages are distinguished. The second multi-dimensional detection dataset is obtained after dimension alignment and format normalization. Then, the process dimension cross-validation correction is performed on it through a multi-parameter coupled judgment model to filter out abnormal parameters that do not match the logic of the assembly and adjustment process and complete global deviation calibration to obtain the third multi-dimensional effective detection dataset. A two-dimensional feature matching model is constructed based on the third multi-dimensional effective detection dataset. The coupling sensitivity coefficients of geometric and performance parameters are introduced, and the dataset is substituted to perform coupling analysis and calculate the coupling accuracy value. The coupling accuracy value is compared with a preset threshold. If the threshold is met, a basic detection report is generated. If the threshold is not met, deviation data is extracted from the third multi-dimensional effective detection dataset, the deviation magnitude of each deviation data is calculated, and the coupling influence weight corresponding to each deviation data is calculated and output based on the coupling sensitivity coefficient. Based on deviation data, deviation magnitude, coupling influence weight and preset process traceability rules, the causes of core assembly and adjustment processes and calibration parameters are reversed, the assembly and adjustment process of the motor assembly is simulated and iterated to correct, the data with remaining deviations are coupled and compensated to correct, and the fourth multi-dimensional detection dataset is obtained. The fourth multi-dimensional detection dataset is used to back-substitute the model to recalculate the coupling accuracy value. The effectiveness of the correction is verified based on the change in accuracy value between the two tests. The entire process of detection data, causes of deviation, correction measures and verification results are integrated to generate a test report, thereby realizing the detection of the drive motor assembly.

2. The testing method for the new energy vehicle drive motor assembly as described in claim 1, characterized in that, The process involves acquiring geometric and performance characteristic data of the drive motor assembly, and then using a virtual parameter fitting algorithm based on the dynamic correlation between assembly and adjustment processes and parameters. This involves fusing actual assembly and adjustment process operating condition information with standard range matching, interpolation completion, deviation calibration, and range constraint fitting to establish a first multi-dimensional detection dataset. This includes: acquiring geometric and performance characteristic data through full-size 3D modeling and scanning of the drive motor assembly; collecting actual assembly and adjustment process operating condition information such as assembly tooling pressure, machining speed, and feed rate; fusing the characteristic data with operating condition information and completing standard range interval matching mapping based on the virtual parameter fitting algorithm based on the dynamic correlation between assembly and adjustment processes and parameters; and sequentially performing numerical interpolation completion, system deviation calibration, and range boundary constraint fitting on the fused data to form a structured first multi-dimensional detection dataset.

3. The testing method for the new energy vehicle drive motor assembly as described in claim 2, characterized in that, The process of performing time-series identification of process interference features and invalid data masking on the first multi-dimensional detection dataset, distinguishing interference distortion and logical conflict data from different assembly and adjustment process stages, and obtaining a second multi-dimensional detection dataset after dimension alignment and format normalization includes: performing dimensional feature extraction and clustering on the first multi-dimensional detection dataset according to the assembly and adjustment process stages based on multi-dimensional data interference feature clustering and outlier filtering algorithms, distinguishing regular valid data from interference distortion and logical conflict data from process stages; setting interference judgment thresholds and outlier verification rules related to process stages, marking and masking invalid data entries to obtain a relay multi-dimensional detection dataset; and performing full-dimensional alignment and standardized format normalization on the relay multi-dimensional detection dataset to generate the second multi-dimensional detection dataset.

4. The testing method for the new energy vehicle drive motor assembly as described in claim 3, characterized in that, The process involves cross-validating and correcting the parameters in the second multi-dimensional detection dataset using a multi-parameter coupled judgment model, filtering out abnormal parameters that do not match the assembly and adjustment process logic, and completing global deviation calibration to obtain a third multi-dimensional effective detection dataset. This includes: performing cross-validation of each parameter in the second multi-dimensional detection dataset item by item in the process dimension, verifying the logical compatibility and numerical rationality between parameters in conjunction with the assembly and adjustment process logic, and determining effective parameters; filtering out abnormal parameters that do not match the assembly and adjustment process logic, have logical contradictions, or whose values ​​exceed the standard range, performing full-dimensional global deviation calibration on the effective parameters, and unifying the numerical benchmark and dimension format to obtain the third multi-dimensional effective detection dataset.

5. The testing method for the new energy vehicle drive motor assembly as described in claim 4, characterized in that, The process of constructing a two-dimensional feature matching model based on the third multi-dimensional effective detection dataset, introducing coupling sensitivity coefficients of geometric and performance parameters, substituting them into the dataset for coupling analysis, and calculating the coupling accuracy value includes: constructing a two-dimensional feature matching model with assembly accuracy feature analysis branch, operational performance feature analysis branch, and coupling accuracy quantification layer based on the parameter dimensions and feature attributes of the third multi-dimensional effective detection dataset; introducing coupling sensitivity coefficients of geometric and performance parameters into the coupling accuracy quantification layer, setting parameter coupling correlation quantification rules; importing the third multi-dimensional effective detection dataset into the model for multi-dimensional parameter feature matching and correlation coupling analysis, combining the coupling sensitivity coefficients to quantify the degree of collaborative adaptation between parameters, and outputting the coupling accuracy value.

6. The testing method for the new energy vehicle drive motor assembly as described in claim 5, characterized in that, The process involves inputting the dataset for coupling analysis and calculating the coupling accuracy value. This is then compared to a preset threshold. If the threshold is met, a basic detection report is generated; otherwise, valid deviation detection data and the coupling influence weights of the deviation parameters are output. This includes: comparing the calculated coupling accuracy value with a preset acceptable threshold; if the threshold requirement is met, a basic detection report containing core detection parameters is generated; if the threshold requirement is not met, deviation parameters are extracted and located from the third multi-dimensional valid detection dataset, valid deviation detection data is output, and the coupling influence weights of each deviation parameter are calculated and output based on the coupling sensitivity coefficient.

7. The testing method for the new energy vehicle drive motor assembly as described in claim 6, characterized in that, The process involves: using deviation data, deviation amplitude, coupling influence weights, and preset process tracing rules to reverse-engineer the causes and calibration parameters of core assembly and adjustment processes; performing simulation-based iterative corrections on the motor assembly assembly process; and applying coupling compensation corrections to data with remaining deviations to obtain a fourth multi-dimensional detection dataset. This includes: decomposing valid deviation detection data into single-dimensional independent parameter items; combining deviation amplitude and coupling influence weights to match the correlation characteristics between parameters and assembly and adjustment processes; and reverse-engineering the nodes and causes of core assembly and adjustment processes based on preset process tracing rules. It also involves calculating standardized process calibration parameters corresponding to the causes of core assembly and adjustment processes based on preset judgment rules and process characteristics; performing simulation-based iterative corrections on the motor assembly assembly assembly process based on calibration parameters; simulating multiple rounds of process parameter adjustments and verifying the effects to determine valid detection data with remaining deviations; synchronously adjusting associated geometric and performance characteristic data according to coupling influence weights; applying coupling compensation corrections to data with remaining deviations to calibrate parameter values ​​to the standard range; and merging and integrating the detection data with completed deviation compensation with the original qualified detection data to obtain the fourth multi-dimensional detection dataset.

8. The testing method for the new energy vehicle drive motor assembly as described in claim 7, characterized in that, The process involves back-substituting the fourth multi-dimensional detection dataset into the model to recalculate the coupling accuracy value, verifying the effectiveness of the correction based on the change in accuracy value between the two calculations, and integrating the entire process of detection data, causes of deviation, correction measures, and verification results to generate a detection report, including: Substitute the fourth multi-dimensional detection dataset back into the two-dimensional feature matching model, and recalculate the coupling accuracy value again using the repeatability accuracy calculation process; calculate the change in the coupling accuracy value between the two calculations, compare it with the preset correction effectiveness threshold, and verify the actual effectiveness of simulation iterative correction and coupling compensation correction; integrate the full-process detection data, deviation causes, simulation iterative correction steps, coupling compensation correction parameters, and correction effectiveness verification results, and generate a detection report containing full-dimensional detection information in a standardized format.

9. A drive motor assembly testing apparatus for implementing the drive motor assembly testing method according to any one of claims 1 to 8, characterized in that, include: The system consists of three parts: model extraction, parameter processing, and detection result generation.

10. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored on the memory, wherein the processor is configured to run the computer program to perform the drive motor assembly detection method according to any one of claims 1 to 8.

Images