Motor drive production test fixture system

Abstract

The application provides a motor drive production test tool system, belonging to the technical field of industrial automation, comprising: an automatic test module for task testing of production equipment according to a test flow formed by a plurality of test instances, and dynamic adjustment of the test instances based on real-time test data; a data analysis module for local correlation analysis and global analysis of task test results according to test indexes of corresponding test instances, and causal relationship analysis across test instances through a Bayesian network to generate an initial report containing a fault root cause tracing conclusion; and an interface interaction module for receiving each test instruction input by a user to perform specified testing on the production equipment, and generating a test report according to matching content of each specified test result and the initial report, in combination with test permissions of each test instruction input by the user. The problems of low efficiency and insufficient accuracy of traditional testing are solved.

Description

Technical Field

[0001] This invention relates to the field of industrial automation technology, and in particular to a motor drive production testing fixture system. Background Technology

[0002] In the current field of production equipment testing, traditional testing methods mainly rely on manual operation, which is not only inefficient but also prone to false positives and false negatives. Furthermore, traditional methods often fail to provide comprehensive and accurate testing of the correctness of software programs, input / output signals, and operating parameters.

[0003] Therefore, this invention proposes a production testing fixture system for motor drives. Summary of the Invention

[0004] This invention provides a motor drive production testing fixture system to solve the aforementioned technical problems.

[0005] This invention provides a production testing fixture system for electric motor drives, comprising:

[0006] An automated testing module is used to perform task testing on production equipment based on a test process composed of several test instances, and to dynamically adjust the test instances based on real-time test data. When any parameter in the real-time test data fluctuates beyond the corresponding preset range, a verification test item associated with the corresponding parameter is automatically added. The several test instances include: software test instances of production equipment, signal test instances of input and output of production equipment, and test instances of working parameters of production equipment.

[0007] The data analysis module is used to perform local correlation analysis and global analysis on the task test results according to the test indicators of the corresponding test instances, as well as to perform causal relationship analysis across test instances through Bayesian networks, and generate an initial report containing the conclusions of fault root cause tracing.

[0008] The interface interaction module is used to receive various test instructions input by the user to perform specified tests on the production equipment, and generate a test report based on the matching content of each specified test result and the initial report, combined with the test permissions of each test instruction input by the user. Specifically, when the matching content is an empty set, a new panel is created; when the matching content is not an empty set, an index label consistent with the corresponding test instruction is created based on the matching content, and an optimized test report is generated for the user to view.

[0009] Preferably, the automated testing module includes:

[0010] The fault acquisition unit is used to retrieve the historical fault information of the production equipment from the historical equipment working library, match the faults of each production component with the equipment structure diagram of the production equipment, and generate a working condition correction factor by combining the real-time working condition parameters of the production equipment to correct the weight of each fault in the fault set.

[0011] The condition determination unit is used to determine the test objectives for each production component based on software testing according to the corrected fault set, and to determine the triggering conditions based on each test objective.

[0012] The sub-instance acquisition unit is used to retrieve test items that match the triggering conditions from the software instance database, respond to all test identifiers of each test item and test scripts of each test identifier, and configure a startup file for each test item based on the logical topic of each test script and the logical association of test scripts under different test identifiers to obtain a sub-software test instance.

[0013] The vector construction unit is used to determine the fault priority of each production component based on the corrected fault set. It combines the operating priority, guarantee priority, and newly added real-time operating condition dimension parameters of each production component based on the factory theoretical attributes of the equipment structure diagram to construct a priority vector.

[0014] The logic building unit is used to construct a parallel-serial hybrid execution logic line for the production equipment based on the priority vector of each production component, and sort all sub-software test instances according to the parallel-serial hybrid execution logic line to obtain software test instances, wherein high-priority sub-software test instances adopt parallel execution mode and low-priority sub-software test instances adopt serial execution mode.

[0015] Preferably, the fault set includes major fault types, the number of faults based on the same major fault type, the degree of fault based on each fault occurrence, the lower-level refined components of the major fault type to which each fault belongs, and the refined faults of the lower-level components.

[0016] Preferably, the condition determination unit includes:

[0017] The map analysis subunit is used to organize and analyze each fault set according to the fault types involved to obtain a separate classification map;

[0018] The weighting subunit is used to encode the node attributes of each graph line in the separate classification graph, input the working condition-weight coupling model that incorporates real-time load rate parameters, and obtain the dynamic influence weight of each node in the graph line under the current working condition.

[0019] The node identification unit is used to identify the main core nodes and branch-derived nodes of the graph line based on the dynamic influence weight and the fault type hierarchical threshold logic.

[0020] The target determination subunit is used to construct a time-series feature sequence for the main core nodes according to the fault propagation time sequence and extract key mutation points. The fault features corresponding to the key mutation points are used as core initial targets. At the same time, the shortest path is calculated for the association path between the branch derivative nodes and the main core nodes, and the derivative fault features at the end of the path are extracted as derivative initial targets. Each initial target carries a corresponding spectral line traceability identifier and influence weight value.

[0021] The target integration subunit is used to integrate all initial targets involved in the corresponding fault set according to the connection relationship between the main route and branch routes and the main route of the separately classified map. When the initial targets of the main route and branch routes conflict, the target to be retained is determined based on the node influence weight and real-time operating parameters to obtain the first target.

[0022] The target clustering subunit is used to perform cluster analysis on all first targets and automatically adjust the number of clusters based on the density of real-time fault data. When the number of occurrences of any type of fault within a preset period exceeds a threshold, the sub-clusters are automatically split and subdivided to obtain the corresponding cluster targets. The triggering conditions corresponding to the cluster targets are matched from the target-condition lookup table. Each triggering condition is associated with at least one conditional operation expression, and the cluster target is the test target.

[0023] Preferably, the cumulative weight percentage threshold for the main core nodes of electrical faults is set to a1, and the cumulative weight percentage threshold for the main core nodes of mechanical faults is set to a2. The main core nodes are a continuous sequence of nodes in the graph that meets the corresponding threshold, and the branch-derived nodes are related nodes outside the continuous node sequence.

[0024] Preferably, the logical building block includes:

[0025] The extended sub-unit is used to dynamically expand the priority vector of production components based on the newly added real-time resource utilization dimension and test item correlation dimension;

[0026] The conflict detection subunit is used to perform real-time resource conflict detection on high-priority sub-software test instances. If the hardware resources required by high-priority instances do not overlap, independent resource channels are allocated to achieve parallel execution. If resource overlap exists, clustering is performed based on the correlation of test items, and high-priority instances with a correlation higher than a preset threshold are merged into resource-sharing groups. Within each group, execution is performed serially according to dependencies, and execution is performed in parallel between groups. For low-priority sub-software test instances, an acyclic dependency graph is constructed based on the correlation of test items, and the instances are divided into several serial execution sequences according to topological sorting. The start triggering condition of each sequence is bound to the execution progress of high-priority instances.

[0027] The real-time monitoring subunit is used to monitor the test time fluctuations of high-priority instances that are executed in parallel in real time. When the test time of any high-priority instance exceeds the preset proportion of the historical average of the same type, the subsequent low-priority instances that have not been executed are automatically switched to the preload state and executed immediately after the resources are released.

[0028] Preferably, the data analysis module includes:

[0029] The error analysis unit is used to determine the reference result distribution of the corresponding test indicators under the preset indicator dimensions based on the task test results and the test indicators of each test instance. It also combines the single test instance result distribution of the task test result subset of each test instance to construct the local error sequence of the corresponding test instance under the relevant test indicators, and integrates all layout error sequences to establish a global error list across test instances.

[0030] The causal analysis unit is used to solve the global error list based on a Bayesian network to obtain an analysis list based on each test index, identify the first element with significant causal relationship in each analysis list, determine whether the number of the first element is less than a preset number threshold, if it is less, remove redundant elements without causal relationship in the analysis list, and associate the causal relationships of the remaining elements to obtain a causal set.

[0031] Otherwise, based on max (the minimum number of rows required by the preset quantity threshold, and the number of rows in the analysis list occupied by the first element), the corresponding analysis list is decomposed to determine the causal relationship of each sub-table and obtain the causal set.

[0032] The report generation unit is used to correlate and fuse the causal sets of all test metrics to obtain an initial report.

[0033] Preferably, the interface interaction module includes:

[0034] The index building unit is used to receive test instructions input by the user, synchronously parse the test object identifier, test parameter range and user permission level contained in the instruction, and build an instruction-report mapping index based on the test object identifier;

[0035] The classification processing unit is used to automatically generate a new structured panel based on the test parameter range if the matched content is an empty set. The new panel includes the theoretical parameter baseline of the test object, the dynamic drawing area of ​​the real-time test curve, and the associated historical test data comparison column of the same type of equipment.

[0036] If the matched content is not an empty set, cluster analysis is performed on the matched content to obtain several association clusters. Each association cluster corresponds to a different parameter dimension in the test instruction. Based on the weight of the association cluster, a multi-level index label consistent with the test instruction is established. The first-level label corresponds to the major category of the test object, and the second-level label corresponds to the specific parameter item.

[0037] The report synthesis unit is used to generate test reports based on the processing results of the matched content and in conjunction with the instruction-report mapping index library.

[0038] Compared with the prior art, the beneficial effects of this application are as follows:

[0039] The automated testing module enables unmanned and dynamically adjusted testing processes, avoiding the inefficiency and inaccuracy of manual operations. The data analysis module driven by Bayesian networks breaks down information barriers between test instances, accurately tracing the root cause of faults. The interface interaction module driven by permission adaptation and matching logic generates personalized reports to meet the information needs of different users, thereby improving the overall efficiency, accuracy, and fault diagnosis efficiency of motor drive equipment production testing.

[0040] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0041] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0042] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0043] Figure 1 As described in the embodiments of the present invention Detailed Implementation

[0044] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0045] This invention provides a production testing fixture system for motor drives, such as... Figure 1 As shown, it includes:

[0046] An automated testing module is used to perform task testing on production equipment based on a test process composed of several test instances, and to dynamically adjust the test instances based on real-time test data. When any parameter in the real-time test data fluctuates beyond the corresponding preset range, a verification test item associated with the corresponding parameter is automatically added. The several test instances include: software test instances of production equipment, signal test instances of input and output of production equipment, and test instances of working parameters of production equipment.

[0047] The data analysis module is used to perform local correlation analysis and global analysis on the task test results according to the test indicators of the corresponding test instances, as well as to perform causal relationship analysis across test instances through Bayesian networks, and generate an initial report containing the conclusions of fault root cause tracing.

[0048] The interface interaction module is used to receive various test instructions input by the user to perform specified tests on the production equipment, and generate a test report based on the matching content of each specified test result and the initial report, combined with the test permissions of each test instruction input by the user. Specifically, when the matching content is an empty set, a new panel is created; when the matching content is not an empty set, an index label consistent with the corresponding test instruction is created based on the matching content, and an optimized test report is generated for the user to view.

[0049] In this embodiment, the test instance is a predefined standardized test content for a specific test dimension of the motor drive equipment. Test instance configuration files are written in XML format and stored in the system database. Each instance includes fields such as test ID, test type, parameter range, and execution script path. For example, a software test instance has the ID SW-001, the test type is motor start-stop logic test, the parameter standard is start-stop response time ≤ 0.5s, and the execution script is Motor_StartStop.exe; an input signal test instance has the ID SI-002, the test type is emergency stop signal test, the parameter standard is device power-off time ≤ 0.2s after emergency stop signal input, and the execution script is Emergency_Signal.bat; and a working parameter test instance has the ID WP-003, the test type is motor speed test, the parameter standard is speed deviation ±2% under rated load, and the execution script is Speed_Test.py.

[0050] In this embodiment, the real-time test data dynamic adjustment test example is that the system collects equipment parameters (such as current, voltage, and speed) in real time during the test process. If any parameter exceeds the preset range, a verification test item associated with that parameter is automatically added without manual intervention. 1. Real-time parameters of the equipment are collected through a PLC (Programmable Logic Controller), with the sampling frequency set to 10Hz; 2. A parameter monitoring program is deployed in the automated test module to compare the real-time parameters with the preset range (such as the preset range of 5-10A for motor current); 3. If the parameter exceeds the range, the test item addition logic is triggered, and the associated verification test item is called from the database. In the test, the real-time value of the motor current is 12A (exceeding the 5-10A range), and the system automatically adds a current stability verification test item (IDWP-004). The test content is to collect 20 sets of current data under 80%, 100%, and 120% load and determine whether the fluctuation is ≤1A.

[0051] In this embodiment, a Bayesian network is a graphical model based on probabilistic reasoning. Here, it is used to establish causal relationships between test results of different test instances (software, signals, parameters), calculate the probability that an abnormal test result leads to an abnormal test result, and locate the root cause of the fault. 1. Construct a Bayesian network model: nodes are the results of each test instance (such as software response delay, signal transmission delay, current anomaly), and edges are the causal relationship probabilities between nodes (trained based on historical fault data); 2. Input the current test results (such as software response delay = 0.8s (abnormal), signal transmission delay = 0.3s (abnormal)), and use the PyMC3 tool to calculate the posterior probability of each node to determine the causal relationship.

[0052] In this embodiment, test permissions are based on report viewing permissions set according to user roles (such as operator, engineer, and administrator). Different roles can view different report contents (such as parameter compliance and fault tracing conclusions). Specifically, a user role-permission mapping table is established in the system backend (e.g., operator: only view parameter test results; engineer: view all test results and fault tracing; administrator: additionally view system configuration). After receiving user login information, the interface interaction module calls the permission verification interface to filter out content in the report that the user does not have permission to view. For example, after logging in, the operator can view the WP-003 test report and it only shows a speed deviation of 1.5% (qualified); after logging in, the engineer can view the causal analysis of the speed deviation in addition to the parameter results: no abnormal correlation; after logging in, the administrator can additionally display the test script version: V2.1.

[0053] In this embodiment, an empty set is, for example, a user-specified test of the inverter cooling fan speed (without an initial report). The system creates a new panel that includes the theoretical fan speed value (2000 rpm), a real-time curve area, and a comparison column of historical test data for the same model of fan.

[0054] For non-empty sets, such as user-specified motor speed tests (which are included in the initial report), the system establishes index tags: Level 1: Motor parameter test; Level 2: Rated load speed. Clicking the tag will directly locate the speed test results in the initial report.

[0055] The beneficial effects of the above technical solution are as follows: The automated testing module enables unmanned and dynamically adjusted testing processes, avoiding the inefficiency and inaccuracy of manual operation; the Bayesian network-driven data analysis module breaks down information barriers between test instances, accurately tracing the root cause of faults; and the permission adaptation and matching logic-driven interface interaction module generates personalized reports to meet the information needs of different users, comprehensively improving the efficiency, accuracy, and fault diagnosis efficiency of motor drive equipment production testing. The testing time of this invention is reduced by 60% compared to traditional manual testing and by 25% compared to existing automated systems. The fault root cause tracing accuracy of this invention reaches 92%, which is 18% higher than existing Bayesian network diagnostic methods. The missed detection rate of this invention is less than 0.5%, while the missed detection rate of traditional manual testing is 8%.

[0056] This invention provides a production testing fixture system for electric motor drives, wherein the automated testing module includes:

[0057] The fault acquisition unit is used to retrieve the historical fault information of the production equipment from the historical equipment working library, match the faults of each production component with the equipment structure diagram of the production equipment, and generate a working condition correction factor by combining the real-time working condition parameters of the production equipment to correct the weight of each fault in the fault set.

[0058] The condition determination unit is used to determine the test objectives for each production component based on software testing according to the corrected fault set, and to determine the triggering conditions based on each test objective.

[0059] The sub-instance acquisition unit is used to retrieve test items that match the triggering conditions from the software instance database, respond to all test identifiers of each test item and test scripts of each test identifier, and configure a startup file for each test item based on the logical topic of each test script and the logical association of test scripts under different test identifiers to obtain a sub-software test instance.

[0060] The vector construction unit is used to determine the fault priority of each production component based on the corrected fault set. It combines the operating priority, guarantee priority, and newly added real-time operating condition dimension parameters of each production component based on the factory theoretical attributes of the equipment structure diagram to construct a priority vector.

[0061] The logic building unit is used to construct a parallel-serial hybrid execution logic line for the production equipment based on the priority vector of each production component, and sort all sub-software test instances according to the parallel-serial hybrid execution logic line to obtain software test instances, wherein high-priority sub-software test instances adopt parallel execution mode and low-priority sub-software test instances adopt serial execution mode.

[0062] Preferably, the fault set includes major fault types, the number of faults based on the same major fault type, the degree of fault based on each fault occurrence, the lower-level refined components of the major fault type to which each fault belongs, and the refined faults of the lower-level components.

[0063] In this embodiment, the historical equipment working database stores historical fault data of motor drive equipment, including fields such as fault occurrence time, component, phenomenon, and handling solution, which are used to extract fault patterns. Specifically: 1. A historical equipment working database is built using a MySQL database. The data table fields include equipment number, fault time, fault component, fault type, fault severity, and handling record; 2. The fault acquisition unit retrieves the fault data of the target equipment through SQL statements (such as SELECT * FROM fault WHERE device_id='M001'). For example, retrieving the fault data of the frequency converter with equipment number M001 yields 15 records, including 8 overcurrent faults, 3 short circuit faults, and 4 overheating faults. The fault times are concentrated in summer (ambient temperature > 30℃).

[0064] In this embodiment, the equipment structure diagram is a mechanical and electrical assembly diagram of the motor drive system, marking the physical location and electrical connection relationship of each component (motor, frequency converter, sensor) for fault location matching. Specifically, the equipment structure diagram is drawn using AutoCAD, exported as SVG format and stored in the system; a unique identifier ID (e.g., frequency converter ID: INV-001) is added to each component in the structure diagram. The fault acquisition unit locates the fault location by matching the fault component ID with the identifier in the structure diagram. For example, in the fault data, the fault component = frequency converter (ID: INV-001), and the system highlights the location of INV-001 in the equipment structure diagram (located in the middle of the left side of the equipment, connecting the motor power line and the sensor signal line).

[0065] In this embodiment, the operating condition correction factor ,in, This is the real-time load rate, ranging from 0 to 1, reflecting the current load intensity of the equipment; This is the real-time temperature deviation, which is the difference between the current temperature and the normal operating temperature, and is greater than 0. This represents the maximum temperature deviation during equipment operation, typically 20℃. To accumulate runtime; The design lifespan of the equipment is set at the factory. This represents the frequency of occurrence of historical fault type i0; n1 represents the impact of the i0th type of historical fault, with a value ranging from 0 to 1. The higher the impact, the more severe the consequences. n1 represents the total number of historical fault types. It is based on multiple linear regression of 500 sets of historical fault-operating condition data, ensuring that the coefficients allow the correction factor to fluctuate reasonably between 1 and 3.

[0066] In this embodiment, ,in, The real-time weights after correction for the i0th type of fault. The original historical weights for the i0th type of fault are obtained based on historical occurrence frequency statistics. This represents the impact of the i0th type of fault, with a value ranging from 0 to 1. A higher impact indicates a more severe fault consequence.

[0067] ,in, The fault priority of the j0th production component, The real-time weight after correction for the fault type of the j0th production component; The running priority of the j0th production component; The priority for ensuring the production of the j0th component; The weighting coefficients were determined based on multiple linear regression analysis of 300 fault handling cases. =1, with values ​​of 0.3, 0.3, 0.2, and 0.2 respectively.

[0068] In this embodiment, the software instance database is a database that stores preset software test items, including test items, test identifiers, test scripts, startup file configuration rules, and is used to match test items corresponding to trigger conditions.

[0069] In this embodiment, the priority vector comprises a four-dimensional vector of fault priority, operation priority, protection priority, and real-time operating parameters, used to characterize the testing priority of production components. Specifically: Fault priority: determined based on the corrected fault weight; Operation priority: core components (such as frequency converters) are set to 0.9, and auxiliary components (such as cooling fans) are set to 0.5; Protection priority: components affecting equipment safety (such as emergency stop modules) are set to 0.9, and non-safety components are set to 0.4; Real-time operating parameters: determined based on load rate normalization; Priority vector = (fault priority, operation priority, protection priority, real-time operating parameters).

[0070] In this embodiment, the parallel-serial hybrid execution logic line determines the execution order of test instances based on priority vectors. High-priority instances are executed in parallel (occupying different resources simultaneously), while low-priority instances are executed serially (occupying the same resource in sequence). For example, high-priority instances such as inverter overcurrent test (vector mean 0.825) and motor overload test (vector mean 0.78) are executed in parallel using DAQ card channels 1 and 2, respectively. Low-priority instances such as cooling fan test (vector mean 0.45) and indicator light test (vector mean 0.3) are executed serially in the order of fan → indicator light.

[0071] In this embodiment, major fault types are primary classifications of faults, based on the system dimension of the fault occurrence, reflecting the overall attributes of the fault. For example, electrical faults (overcurrent, short circuit, leakage), mechanical faults (bearing wear, shaft misalignment, loose components), and software faults (logic loopholes, response delays). For example, fault phenomenon E01 (overcurrent) alarm code of the frequency converter → classified as an electrical fault; fault phenomenon abnormal noise of the motor and excessive vibration value → classified as a mechanical fault; fault phenomenon no response of the equipment after the start / stop command is issued → classified as a software fault.

[0072] In this embodiment, the number of failures of the same major fault type is the total number of failures of the same major fault type that occur in a certain device within a historical period (e.g., 3 years), reflecting the frequency of such failures. For example, in 20 years, device M0013 experienced 20 electrical failures (8 overcurrents, 5 short circuits, and 7 leakages), 12 mechanical failures (6 bearing wears, 3 shaft misalignments, and 3 loose components), and 8 software failures.

[0073] In this embodiment, the severity of each fault is determined by classifying the severity of a single fault based on its impact range, repair cost, and downtime (mild, moderate, severe). Specifically, the fault severity classification standards are preset as follows: Mild: alarm only, no downtime, automatic recovery (e.g., instantaneous overcurrent, recovers within 1 second); Moderate: manual reset required, downtime < 1 hour (e.g., continuous overcurrent, manual power-off reset required); Severe: component damage, parts need to be replaced, downtime > 1 hour (e.g., IGBT module burnout). The fault acquisition unit matches the fault severity according to the processing records. For example, out of 8 overcurrent faults, 3 are mild (instantaneous overcurrent, automatic recovery), 4 are moderate (manual reset, 30-minute downtime), and 1 is severe (IGBT burnout, 4-hour downtime, parts replacement required).

[0074] In this embodiment, the lower-level refined components are the hierarchical relationship of second-level components → third-level components corresponding to major fault types, reflecting the specific component location where the fault occurred. Specifically, a component hierarchy is established in the equipment structure diagram (e.g., electrical fault → inverter → IGBT module; mechanical fault → motor → bearing). The fault acquisition unit traces the lower-level refined components based on the fault component ID. For example, electrical fault → refined component is inverter (second-level) → IGBT module (third-level); mechanical fault → refined component is motor (second-level) → bearing (third-level).

[0075] In this embodiment, the detailed faults of lower-level components are the specific fault manifestations of those components, distinct from the macroscopic classification of major fault types. They reflect the specific causes of the faults and are specifically defined as follows: a pre-defined mapping relationship between detailed faults and components is established (e.g., IGBT module → detailed faults: short circuit, open circuit, overheating; bearing → detailed faults: wear, insufficient lubrication, abnormal noise). The fault acquisition unit matches detailed faults based on the fault phenomena. For example, in the detailed faults of the IGBT module: out of 8 overcurrent faults, 5 were IGBT short circuits and 3 were IGBT overheating; in the detailed faults of the bearing: out of 6 wear faults, 4 were wear due to insufficient lubrication and 2 were normal aging wear.

[0076] The beneficial effects of the above technical solution are: by integrating historical faults and real-time operating conditions, an accurate fault set is generated; by matching target test items, the test is ensured to be targeted; by optimizing the test execution order, high-priority tests are performed in parallel and low-priority tests are performed sequentially, thereby improving test efficiency and prioritizing the coverage of high-risk test items to reduce the risk of missed tests.

[0077] This invention provides a production testing fixture system for electric motor drives, wherein the condition determination unit includes:

[0078] The map analysis subunit is used to organize and analyze each fault set according to the fault types involved to obtain a separate classification map;

[0079] The weighting subunit is used to encode the node attributes of each graph line in the separate classification graph, input the working condition-weight coupling model that incorporates real-time load rate parameters, and obtain the dynamic influence weight of each node in the graph line under the current working condition.

[0080] The node identification unit is used to identify the main core nodes and branch-derived nodes of the graph line based on the dynamic influence weight and the fault type hierarchical threshold logic.

[0081] The target determination subunit is used to construct a time-series feature sequence for the main core nodes according to the fault propagation time sequence and extract key mutation points. The fault features corresponding to the key mutation points are used as core initial targets. At the same time, the shortest path is calculated for the association path between the branch derivative nodes and the main core nodes, and the derivative fault features at the end of the path are extracted as derivative initial targets. Each initial target carries a corresponding spectral line traceability identifier and influence weight value.

[0082] The target integration subunit is used to integrate all initial targets involved in the corresponding fault set according to the connection relationship between the main route and branch routes and the main route of the separately classified map. When the initial targets of the main route and branch routes conflict, the target to be retained is determined based on the node influence weight and real-time operating parameters to obtain the first target.

[0083] The target clustering subunit is used to perform cluster analysis on all first targets and automatically adjust the number of clusters based on the density of real-time fault data. When the number of occurrences of any type of fault within a preset period exceeds a threshold, the sub-clusters are automatically split and subdivided to obtain the corresponding cluster targets. The triggering conditions corresponding to the cluster targets are matched from the target-condition lookup table. Each triggering condition is associated with at least one conditional operation expression, and the cluster target is the test target.

[0084] Preferably, the cumulative weight percentage threshold for the main core nodes of electrical faults is set to a1, and the cumulative weight percentage threshold for the main core nodes of mechanical faults is set to a2. The main core nodes are a continuous sequence of nodes in the graph that meets the corresponding threshold, and the branch-derived nodes are related nodes outside the continuous node sequence.

[0085] In this embodiment, the separate classification graph is a directed graph constructed for a single fault set (such as the inverter electrical fault set) according to the hierarchical relationship of major fault type → secondary component → tertiary component → detailed fault. Each edge represents the association relationship between the component and the fault. Specifically, the graph is constructed using the Graphviz tool, with nodes representing major fault type, component, and detailed fault, and edges representing the association relationship. The graph parsing subunit reads the fault set data, automatically generates the graph, and exports it as a PNG format for viewing. For example, the separate classification graph of the inverter electrical fault set is: node electrical fault (A) → inverter (B) → IGBT module (C) → short circuit (D1) / overheating (D2), with edges A→B, B→C, C→D1, and C→D2.

[0086] In this embodiment, node attribute feature encoding converts the attributes of the graph nodes (fault occurrence probability, impact range, repair difficulty) into numerical vectors (feature vectors) that can be input into the model. Specifically, node attribute quantification is as follows: fault occurrence probability (0-1, calculated based on historical occurrences), impact range (0-1, core component = 1, auxiliary component = 0.5), repair difficulty (0-1, replacement component = 1, reset = 0.3); feature vector = (fault occurrence probability, impact range, repair difficulty), using Min-Max normalization (mapped to 0-1). For example, the attributes of the IGBT module (node ​​C) are: fault occurrence probability = 8 / 20 = 0.4 (8 out of 20 electrical faults are associated with C), impact range = 1 (core component), repair difficulty = 0.9 (replacing the IGBT is costly), and feature vector = (0.4, 1, 0.9).

[0087] In this embodiment, the working condition-weight coupling model is an algorithm model that integrates real-time load rate and node feature vectors. It is used to calculate the dynamic influence weight of a node under the current working condition, which reflects the importance of the node to the test target. The model input is the node feature vector and the real-time load rate, and the model output is the dynamic influence weight = (mean of feature vector × weight coefficient) × (1 + 0.2 × load rate). The model training and calculation are implemented using the Scikit-learn library of Python. The weight coefficient is obtained by training with 100 sets of historical fault data and the value is between 0.5 and 2. It is adjusted according to the actual data training results.

[0088] In this embodiment, the fault type hierarchical threshold logic sets different cumulative thresholds for dynamic influence weights for different fault types (electrical and mechanical) to identify the main core nodes (with accumulated weights meeting the standard) and branch-derived nodes (with accumulated weights not meeting the standard) in the graph. Specifically, in the node identification unit, a threshold is preset: electrical type a1 = 70% (0.7), mechanical type a2 = 60% (0.6); for each graph line (such as A→B→C→D1), the dynamic influence weights of the nodes are accumulated sequentially, and the first threshold is reached. The continuous node sequence reaching the threshold is the main core node. For example, the node weights of the spectrum line A→B→C→D1 are: A=0.2, B=0.3, C=0.895, D1=0.1; the cumulative A+B=0.5<0.7, A+B+C=1.395≥0.7, so the main core node is A→B→C, and the branch derived node is D1. Among them, a1=70% is obtained based on the statistics of 200 electrical fault data, and a2=60% is obtained based on the statistics of 150 mechanical fault data.

[0089] In this embodiment, the time-series prediction model takes the fault occurrence time sequence as input (e.g., B occurs 0.5s after A, and C occurs 0.3s after B) and outputs the fault propagation time-series feature sequence; the fault tree model takes equipment failure as the top event and A, B, C, and D1 as bottom events, and calculates the importance of each bottom event; the fusion algorithm extracts key abrupt change points in the time-series feature sequence (e.g., the impact of the fault increases sharply after C occurs), combines them with the fault tree importance, and determines the initial target, such as the time-series feature sequence:

[0090] A(0s)→B(0.5s)→C(0.8s)→D1(1.2s), the key mutation point is C (0.8s, the impact of the fault increases from 0.3 to 0.9), so the core initial target is C (IGBT module fault); the shortest path between the branch derived node D1 and C is C→D1, so the derived initial target is D1 (IGBT short circuit). Both carry the spectral line traceability identifier spectral IDG001 and weights 0.895 / 0.1.

[0091] In this embodiment, the target conflict resolution is to determine the priority target to be retained when there is a contradiction between the initial targets of the trunk and the branch routes (such as the trunk target requires test C, the branch target requires test D1, but the resources can only support one). The conflict resolution formula is: priority score = node influence weight × (1 + 0.1 × real-time load rate).

[0092] In this embodiment, adjusting the number of clusters based on real-time fault data density uses the DBSCAN clustering algorithm. The algorithm automatically adjusts the number of clusters based on the density of real-time fault data, i.e., the number of fault occurrences per unit time, avoiding clustering bias caused by a fixed number of clusters. The DBSCAN algorithm parameters are: neighborhood radius of 0.2 (based on the Euclidean distance of the fault feature vector), and minimum sample size MinPts = 3. If the number of fault occurrences within a cluster exceeds a threshold (e.g., 5 times) within one month, it is automatically split into two sub-clusters. The first target includes IGBT module faults (C), inverter faults (B), and motor bearing faults (E). After clustering, two clusters are formed: Cluster 1 (C, B, electrical faults)... The first target (IGBT module fault, occurring 8 times or more than 5 times in 1 month) is automatically split into two sub-clusters: C (IGBT fault) and B (other inverter faults). Cluster 2 (E, mechanical fault, occurring 3 times in 1 month) remains unchanged. For example, the first target includes IGBT module faults (C), inverter faults (B), and motor bearing faults (E). After clustering, two clusters are formed: Cluster 1 (C, B, electrical faults, occurring 8 times or more than 5 times in 1 month) is automatically split into two sub-clusters: C (IGBT fault) and B (other inverter faults). Cluster 2 (E, mechanical fault, occurring 3 times in 1 month) remains unchanged. The threshold of 5 times is based on historical data from the past year, and 5 times is the critical value for frequent fault occurrence.

[0093] In this embodiment, the target-condition lookup table stores the correspondence between cluster targets and test trigger conditions, and is used to match trigger conditions. The lookup table is stored in Excel. For example, cluster target C (IGBT module failure) matches the record in the lookup table where the target description is IGBT module failure. The trigger condition is that the simulated IGBT input current is 120% of the rated current. The condition operation expression is a logical expression such as: current > 10A and temperature > 50℃.

[0094] In this embodiment, the cumulative weight percentage threshold a1 for electrical faults is the lower limit of the cumulative weight percentage of the dynamic influence of the core nodes in the electrical fault spectrum. A continuous node sequence with a cumulative percentage ≥ a1 is a core node. In the configuration file of the node identification unit, a1 = 0.7 (70%) is set. For example, the total weight of the electrical fault spectrum A→B→C→D1 is: 0.2 + 0.3 + 0.895 + 0.1 = 1.495; the cumulative weight of A + B + C is 1.395, and the cumulative weight percentage is 1.395 / 1.495 ≈ 93.3% ≥ 70%. Therefore, the core node in the main spectrum is A→B→C.

[0095] In this embodiment, the cumulative weight percentage threshold a2 for mechanical faults is the lower limit of the cumulative weight percentage of the dynamic influence of the core nodes in the mechanical fault spectrum. A continuous node sequence with a cumulative weight percentage ≥ a2 is a core node. In the configuration file of the node identification unit, a2 = 0.6 (60%) is set.

[0096] In this embodiment, the main core nodes are a continuous sequence of nodes that meet the threshold, reflecting the main path of fault propagation; the branch-derived nodes are related nodes outside the sequence, reflecting the secondary path of fault propagation. For example, in the electrical type graph, A→B→C is the core node and D1 is the branch-derived node; in the mechanical type graph, F→G→H is the core node and I is the branch-derived node.

[0097] The beneficial effects of the above technical solution are as follows: the fault set is transformed into a visual association graph through graph analysis, the importance of nodes is dynamically calculated by combining the working condition-weight coupling model, and core and derived nodes are identified by hierarchical thresholds; the initial target is accurately extracted by integrating time series prediction and fault tree model, the target is optimized by conflict adjudication and density clustering, and finally targeted triggering conditions are matched to ensure that the test target focuses on high-risk faults and the triggering conditions conform to real-time working conditions, thereby improving the accuracy and adaptability of the test.

[0098] This invention provides a motor drive production testing fixture system, with a logic building unit comprising:

[0099] The extended sub-unit is used to dynamically expand the priority vector of production components based on the newly added real-time resource utilization dimension and test item correlation dimension;

[0100] The conflict detection subunit is used to perform real-time resource conflict detection on high-priority sub-software test instances. If the hardware resources required by high-priority instances do not overlap, independent resource channels are allocated to achieve parallel execution. If resource overlap exists, clustering is performed based on the correlation of test items, and high-priority instances with a correlation higher than a preset threshold are merged into resource-sharing groups. Within each group, execution is performed serially according to dependencies, and execution is performed in parallel between groups. For low-priority sub-software test instances, an acyclic dependency graph is constructed based on the correlation of test items, and the instances are divided into several serial execution sequences according to topological sorting. The start triggering condition of each sequence is bound to the execution progress of high-priority instances.

[0101] The real-time monitoring subunit is used to monitor the test time fluctuations of high-priority instances that are executed in parallel in real time. When the test time of any high-priority instance exceeds the preset proportion of the historical average of the same type, the subsequent low-priority instances that have not been executed are automatically switched to the preload state and executed immediately after the resources are released.

[0102] In this embodiment, the expansion adds two dimensions, real-time resource utilization and test item correlation, to the original four-dimensional priority vector, forming a six-dimensional vector that more comprehensively represents the test priority. The expanded priority vector = (fault priority, operation priority, protection priority, real-time operating parameters, resource utilization, correlation). For example, the expanded vector of the inverter overcurrent test instance is (0.8, 0.9, 0.8, 0.8, 0.6, 0.9), all of which are normalized values. It should be noted that the test item correlation is calculated using cosine similarity.

[0103] In this embodiment, if high-priority instances occupy the same hardware resources (such as DAQ card channels or sensors), it is considered a resource conflict, which is resolved through independent channel allocation or resource sharing groups. Low-priority instances are handled through acyclic dependency graphs and topology sorting, specifically: Resource conflict detection: querying hardware resource occupancy status through system resource monitoring interfaces (such as Windows WMI interfaces); High-priority processing: No conflict → allocate independent channels (e.g., channel 1 for overcurrent testing, channel 2 for overload testing); Conflicts → merge instances with correlation > 0.8 into a sharing group (e.g., overcurrent testing and short-circuit testing have a correlation of 0.85, merge). Group 1), with serial execution within the group and parallel execution between groups; Low-priority processing: Dependencies are represented using a directed graph (e.g., heat dissipation test → temperature test), and topological sorting is performed using the Kahn algorithm to divide them into serial sequences. High-priority instances: Overcurrent test (requires channel 1), overload test (requires channel 2) → no conflict, parallel execution; Overcurrent test (requires channel 1), short-circuit test (requires channel 1) → correlation degree 0.85, merged into group 1, with serial execution within the group as overcurrent → short-circuit; Low-priority instances: Heat dissipation test, temperature test, indicator light test → dependency relationship heat dissipation → temperature, topological sorting is heat dissipation → temperature → indicator light, serial execution.

[0104] In this embodiment, the startup time of the low-priority instance depends on the execution progress of the high-priority instance (e.g., the low-priority instance starts after the high-priority instance has completed 50%) to avoid resource waste. The high-priority instance overcurrent test has a total of 10 steps, and the low-priority instance heat dissipation test startup condition is that the overcurrent test progress is ≥ 50% (5 steps are completed). When the overcurrent test completes the 5th step, the system automatically starts the heat dissipation test.

[0105] In this embodiment, the system queries the test log database for the time taken by similar instances and calculates the average (e.g., the historical average for overcurrent testing is 20 seconds). A timer collects the current time taken every second. If the current time taken is greater than the average × 1.1 (e.g., 22 seconds), preloading is triggered. The system loads the scripts (e.g., heat dissipation test.py) and data templates (e.g., Test_Template.xlsx) of low-priority instances into memory. For example, if the current time taken by the overcurrent test is 23 seconds (the historical average is 20 seconds, exceeding the average by 15%), the system loads the scripts and data for the indicator light test of the low-priority instance into memory. After the overcurrent test is completed, the indicator light test starts immediately, saving 2 seconds of loading time.

[0106] The beneficial effects of the above technical solution are: by expanding the priority vector dimension, the execution requirements of test instances can be evaluated more comprehensively; by resource conflict detection and layered processing, the resource contention problem of high-priority instances can be solved, and the disordered execution of low-priority instances can be avoided; by monitoring and preloading the time consumption fluctuation, the test waiting time can be reduced, and the overall resource utilization and execution efficiency of the test process can be improved.

[0107] This invention provides a production testing fixture system for electric motor drives, wherein the data analysis module includes:

[0108] The error analysis unit is used to determine the reference result distribution of the corresponding test indicators under the preset indicator dimensions based on the task test results and the test indicators of each test instance. It also combines the single test instance result distribution of the task test result subset of each test instance to construct the local error sequence of the corresponding test instance under the relevant test indicators, and integrates all layout error sequences to establish a global error list across test instances.

[0109] The causal analysis unit is used to solve the global error list based on a Bayesian network to obtain an analysis list based on each test index, identify the first element with significant causal relationship in each analysis list, determine whether the number of the first element is less than a preset number threshold, if it is less, remove redundant elements without causal relationship in the analysis list, and associate the causal relationships of the remaining elements to obtain a causal set.

[0110] Otherwise, based on max (the minimum number of rows required by the preset quantity threshold, and the number of rows in the analysis list occupied by the first element), the corresponding analysis list is decomposed to determine the causal relationship of each sub-table and obtain the causal set;

[0111] The report generation unit is used to correlate and fuse the causal sets of all test metrics to obtain an initial report.

[0112] In this embodiment, the reference result distribution is extracted from the equipment technical manual and stored as index ID, standard upper limit, and standard lower limit; the single test instance result distribution is generated by the error analysis unit receiving multiple results of the test instance and using MATLAB's histogram function to generate a distribution histogram. For example, the reference result distribution of the motor speed test index (IDSpeed-001) is [98%, 102%] (i.e., deviation ±2%); the five results of the single test instance WP-003 are 99%, 99.5%, 100%, 99.2%, and 99.8%, with the result distribution concentrated between 99% and 100%.

[0113] In this embodiment, the local error sequence is the sequence of error values ​​of each result of a certain test instance under a certain index and the reference standard; the global error list is the integration of the local error sequences of all test instances. For example, the local error sequence of a single test instance WP-003 is [1%, 0.5%, 0%, 0.8%, 0.2%]; the global error list integrates the local error sequences of SW-001 (software response delay error [0.1s, 0.2s]), SI-002 (signal transmission delay error [0.05s, 0.1s]), and WP-003.

[0114] In this embodiment, a Bayesian network is constructed where nodes are error sequences (such as SW error, SI error, WP error) and edges are causal probabilities (trained based on historical data, such as SW error → SI error probability 0.85). A global error list is input, the posterior probability of each node is calculated, and the first element is selected. For example, after inputting the global error list, the Bayesian network calculates the following: SW error (0.2s) → SI error (0.1s) probability 0.88 (first element), SI error → WP error (1%) probability 0.92 (first element), and SW error → WP error probability 0.7 (not the first element).

[0115] In this embodiment, the preset quantity threshold is 2, and the minimum number of rows occupied is 5. If the number of the first element is 1 (<2): remove unrelated elements (such as SW error → WP error), and the causal set is [(SW error, SI error), (SI error, WP error)]. If the number of the first element is 3 (≥2): the number of rows occupied by the first element is 6, max(5,6)=6, and the analysis list is split into 2 sub-tables (6 rows each), and causal sets are generated respectively. For example, if the number of the first element is 2 (≥2), the number of rows occupied by the first element is 4, the minimum number of rows occupied is 5, and max(5,4)=5, the analysis list (10 rows) is split into 2 sub-tables (5 rows each), and the causal set of sub-table 1 is [(SW error, SI error)], and the causal set of sub-table 2 is [(SI error, WP error)].

[0116] In this embodiment, the causal set of all test indicators is integrated to form an initial report containing error phenomenon, causal relationship, and fault root cause. For example, the initial report includes: 3 test cases, SW error 0.2s, SI error 0.1s, WP error 1%; causal relationship diagram showing SW error → SI error → WP error; fault root cause conclusion: software response delay causes signal transmission delay, which in turn causes motor speed deviation.

[0117] The beneficial effects of the above technical solution are: by constructing local and global error data, it provides a foundation for causal analysis; by accurately identifying causal relationships between errors through Bayesian networks, it avoids the subjective bias of traditional analysis; by integrating causal sets, it generates an initial report containing the root causes of faults, providing data support for subsequent fault investigation and equipment optimization, and improving the efficiency and accuracy of fault tracing.

[0118] This invention provides a production and testing fixture system for electric motor drives, wherein the interface interaction module includes:

[0119] The index building unit is used to receive test instructions input by the user, synchronously parse the test object identifier, test parameter range and user permission level contained in the instruction, and build an instruction-report mapping index based on the test object identifier;

[0120] The classification processing unit is used to automatically generate a new structured panel based on the test parameter range if the matched content is an empty set. The new panel includes the theoretical parameter baseline of the test object, the dynamic drawing area of ​​the real-time test curve, and the associated historical test data comparison column of the same type of equipment.

[0121] If the matched content is not an empty set, cluster analysis is performed on the matched content to obtain several association clusters. Each association cluster corresponds to a different parameter dimension in the test instruction. Based on the weight of the association cluster, a multi-level index label consistent with the test instruction is established. The first-level label corresponds to the major category of the test object, and the second-level label corresponds to the specific parameter item.

[0122] The report synthesis unit is used to generate test reports based on the processing results of the matched content and in conjunction with the instruction-report mapping index library.

[0123] In this embodiment, the classification processing unit further includes:

[0124] Calculate the weights of the associated clusters ,in, For parameter importance, This represents the percentage of times a corresponding cluster matches the corresponding test command in historical tests out of the total number of tests. This is a user permission correction factor; when L=1, it corresponds to low permissions. The value is 0.8, and when L=2, it corresponds to medium permissions. The value is 1.0, and L=3 corresponds to high privileges. The value is 1.2; It is the time decay factor, and , This is the attenuation coefficient, with a value of 0.005; The time interval between the current test and historical matches; It is a real-time correction factor, and , This is the normalization coefficient for the current load rate of the production equipment; This is the normalization coefficient for the difference between the current ambient temperature and the rated temperature. Based on the basic weights, and ; To adjust the weights.

[0125] In this embodiment, if the parameter itself is of higher importance, then... It is 0.6. The value is 0.4. If historical matches have higher reference value, then... It is 0.4. If the priority of the two is equal, then the value is 0.6. The values ​​are 0.5, mainly based on the trade-off between the inherent importance of the parameter and the reliability of historical data in the business scenario. The value of P1 ranges from 0.5 to 1, with the core parameter P1=1 and the auxiliary parameter P1=0.5.

[0126] In this embodiment, Depending on the level of importance placed on differences in user role requirements and the impact of production environment fluctuations in the business scenario, for example, user permissions have a more significant impact on testing requirements, in this case, It is 0.3. If the actual production conditions have a more significant impact on the test results, then... It is 0.2. It is 0.3.

[0127] In this embodiment, test command parsing uses regular expressions to extract command fields (e.g., command test device M001 inverter overcurrent parameter (permission: engineer) → identifier M001-INV, parameter range overcurrent, permission engineer); the index uses a Redis database to store mapping relationships (key = test object identifier, value = historical report path). For example, parsing the user command yields the identifier M001-INV, and the value corresponding to this key in the index is D:\Reports\M001_INV_202405.pdf (inverter test report from May 2024).

[0128] In this embodiment, standard values ​​of indicators (such as the inverter overcurrent threshold of 10A) are extracted from the equipment technical manual. The ECharts.js library is used to read test data in real time and plot curves. Historical test data of the same model of equipment (such as the overcurrent test results of nearly 10 devices) are retrieved from the index library. For example, if the user specifies the test inverter cooling fan speed (without an initial report), the new panel includes: a theoretical baseline of 2000 rpm, a real-time curve (showing the fan speed fluctuation of 1980-2020 rpm during the test), and a historical comparison column (average speed of 1995 rpm for 10 devices, with a fluctuation of ±15 rpm).

[0129] In this embodiment, the K-means algorithm is used to cluster the matched content according to the parameter dimension (such as current parameter cluster and voltage parameter cluster); the first-level label corresponds to the cluster cluster (such as inverter electrical parameters), and the second-level label corresponds to the specific parameter (such as overcurrent threshold and short-circuit protection voltage). For example, if the user specifies the overcurrent parameter of the inverter to be tested (with a matching degree of 80% with the initial report), the inverter electrical parameter cluster is obtained by clustering. The first-level label is inverter electrical parameters, and the second-level labels are overcurrent threshold (10A) and overcurrent response time (0.2s). Clicking the second-level label directly locates the test result of the overcurrent parameter in the report.

[0130] In this embodiment, the report template is a preset HTML template, containing user information, test instructions, test results, causal analysis (visible to engineers with permissions), and an index tag area. Report synthesis involves calling index database data, filling the corresponding areas in the template, and generating the final report. For example, a report for an engineer-permitted user includes: User information: Engineer Li XX; Test instruction: M001 Inverter overcurrent test; Test result: (Overcurrent threshold 10A, qualified); Causal analysis: (No correlation error); Index tag: Inverter electrical parameters → Overcurrent threshold. A report for an operator-permitted user only includes test results and index tags, without causal analysis.

[0131] The beneficial effects of the above technical solution are: the index library building unit enables rapid association between test instructions and historical reports, avoiding time-consuming report searches; the classification processing unit provides a dedicated panel for new test results and establishes accurate indexes for existing results, improving report viewing efficiency; and the report synthesis unit adapts to user permissions to generate personalized reports, ensuring that different roles obtain the information they need, improving user experience and operational efficiency.

[0132] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A production testing fixture system for motor drives, characterized in that, include: An automated testing module is used to perform task testing on production equipment based on a test process composed of several test instances, and to dynamically adjust the test instances based on real-time test data. When any parameter in the real-time test data fluctuates beyond the corresponding preset range, a verification test item associated with the corresponding parameter is automatically added. The several test instances include: software test instances of production equipment, signal test instances of input and output of production equipment, and test instances of working parameters of production equipment. The data analysis module is used to perform local correlation analysis and global analysis on the task test results according to the test indicators of the corresponding test instances, as well as to perform causal relationship analysis across test instances through Bayesian networks, and generate an initial report containing the conclusions of fault root cause tracing. The interface interaction module is used to receive various test instructions input by the user to perform specified tests on the production equipment, and generate a test report based on the matching content of each specified test result and the initial report, combined with the test permissions of each test instruction input by the user. When the matching content is an empty set, a new panel is created; when the matching content is not an empty set, an index label consistent with the corresponding test instruction is created based on the matching content, and an optimized test report is generated for the user to view. The automated testing module includes: The fault acquisition unit is used to retrieve the historical fault information of the production equipment from the historical equipment working library, match the faults of each production component with the equipment structure diagram of the production equipment, and generate a working condition correction factor by combining the real-time working condition parameters of the production equipment to correct the weight of each fault in the fault set. The condition determination unit is used to determine the test objectives for each production component based on software testing according to the corrected fault set, and to determine the triggering conditions based on each test objective. The sub-instance acquisition unit is used to retrieve test items that match the triggering conditions from the software instance database, respond to all test identifiers of each test item and test scripts of each test identifier, and configure a startup file for each test item based on the logical topic of each test script and the logical association of test scripts under different test identifiers to obtain a sub-software test instance. The vector construction unit is used to determine the fault priority of each production component based on the corrected fault set. It combines the operating priority, guarantee priority, and newly added real-time operating condition dimension parameters of each production component based on the factory theoretical attributes of the equipment structure diagram to construct a priority vector. The logic building unit is used to construct a parallel-serial hybrid execution logic line for the production equipment based on the priority vector of each production component, and sort all sub-software test instances according to the parallel-serial hybrid execution logic line to obtain software test instances, wherein high-priority sub-software test instances adopt parallel execution mode and low-priority sub-software test instances adopt serial execution mode.

2. The motor drive production and testing fixture system according to claim 1, characterized in that, The fault set includes major fault types, the number of faults based on the same major fault type, the fault severity based on each fault occurrence, the lower-level refined components of the major fault type to which each fault belongs, and the refined faults of the lower-level components.

3. The motor drive production and testing fixture system according to claim 1, characterized in that, The condition determination unit includes: The map analysis subunit is used to organize and analyze each fault set according to the fault types involved to obtain a separate classification map; The weighting subunit is used to encode the node attributes of each graph line in the separate classification graph, input the working condition-weight coupling model that incorporates real-time load rate parameters, and obtain the dynamic influence weight of each node in the graph line under the current working condition. The node identification unit is used to identify the main core nodes and branch-derived nodes of the graph line based on the dynamic influence weight and the fault type hierarchical threshold logic. The target determination subunit is used to construct a time-series feature sequence for the main core nodes according to the fault propagation time sequence and extract key mutation points. The fault features corresponding to the key mutation points are used as core initial targets. At the same time, the shortest path is calculated for the association path between the branch derivative nodes and the main core nodes, and the derivative fault features at the end of the path are extracted as derivative initial targets. Each initial target carries a corresponding spectral line traceability identifier and influence weight value. The target integration subunit is used to integrate all initial targets involved in the corresponding fault set according to the connection relationship between the main route and branch routes and the main route of the separately classified map. When the initial targets of the main route and branch routes conflict, the target to be retained is determined based on the node influence weight and real-time operating parameters to obtain the first target. The target clustering subunit is used to perform cluster analysis on all first targets and automatically adjust the number of clusters based on the density of real-time fault data. When the number of occurrences of any type of fault within a preset period exceeds a threshold, the sub-clusters are automatically split and subdivided to obtain the corresponding cluster targets. The triggering conditions corresponding to the cluster targets are matched from the target-condition lookup table. Each triggering condition is associated with at least one conditional operation expression, and the cluster target is the test target.

4. The motor drive production and testing fixture system according to claim 3, characterized in that, The cumulative weight percentage threshold for the main core nodes of electrical faults is set to a1, and the cumulative weight percentage threshold for the main core nodes of mechanical faults is set to a2. The main core nodes are the continuous node sequences in the graph that meet the corresponding thresholds, and the branch-derived nodes are the related nodes outside the continuous node sequences.

5. The motor drive production and testing fixture system according to claim 1, characterized in that, Logical building blocks, including: The extended sub-unit is used to dynamically expand the priority vector of production components based on the newly added real-time resource utilization dimension and test item correlation dimension; The conflict detection subunit is used to perform real-time resource conflict detection on high-priority sub-software test instances. If the hardware resources required by high-priority instances do not overlap, independent resource channels are allocated to achieve parallel execution. If resource overlap exists, clustering is performed based on the correlation of test items, and high-priority instances with a correlation higher than a preset threshold are merged into resource-sharing groups. Within each group, execution is performed serially according to dependencies, and execution is performed in parallel between groups. For low-priority sub-software test instances, an acyclic dependency graph is constructed based on the correlation of test items, and the instances are divided into several serial execution sequences according to topological sorting. The start triggering condition of each sequence is bound to the execution progress of high-priority instances. The real-time monitoring subunit is used to monitor the test time fluctuations of high-priority instances that are executed in parallel in real time. When the test time of any high-priority instance exceeds the preset proportion of the historical average of the same type, the subsequent low-priority instances that have not been executed are automatically switched to the preload state and executed immediately after the resources are released.

6. The motor drive production and testing fixture system according to claim 1, characterized in that, The data analysis module includes: The error analysis unit is used to determine the reference result distribution of the corresponding test indicators under the preset indicator dimensions based on the task test results and the test indicators of each test instance. It also combines the single test instance result distribution of the task test result subset of each test instance to construct the local error sequence of the corresponding test instance under the relevant test indicators, and integrates all layout error sequences to establish a global error list across test instances. The causal analysis unit is used to solve the global error list based on a Bayesian network to obtain an analysis list based on each test index, identify the first element with significant causal relationship in each analysis list, determine whether the number of the first element is less than a preset number threshold, if it is less, remove redundant elements without causal relationship in the analysis list, and associate the causal relationships of the remaining elements to obtain a causal set. Otherwise, based on max (the minimum number of rows required by the preset quantity threshold, and the number of rows in the analysis list occupied by the first element), the corresponding analysis list is broken down to determine the causal relationship of each sub-table and obtain the causal set; the report generation unit is used to correlate and fuse the causal sets of all test indicators to obtain the initial report.

7. The motor drive production and testing fixture system according to claim 1, characterized in that, The interface interaction module includes: The index building unit is used to receive test instructions input by the user, synchronously parse the test object identifier, test parameter range and user permission level contained in the instruction, and build an instruction-report mapping index based on the test object identifier; The classification processing unit is used to automatically generate a new structured panel based on the test parameter range if the matched content is an empty set. The new panel includes the theoretical parameter baseline of the test object, the dynamic drawing area of ​​the real-time test curve, and the associated historical test data comparison column of the same type of equipment. If the matched content is not an empty set, cluster analysis is performed on the matched content to obtain several association clusters. Each association cluster corresponds to a different parameter dimension in the test instruction. Based on the weight of the association cluster, a multi-level index label consistent with the test instruction is established. The first-level label corresponds to the major category of the test object, and the second-level label corresponds to the specific parameter item. The report synthesis unit is used to generate a test report based on the processing results of the matched content and in combination with the instruction-report mapping index library.

Images