Template learning and abnormality positioning based module connection automatic detection method and system

Abstract

The application provides a module connection automatic detection method and system based on template learning and abnormal positioning, aiming to improve the abnormal positioning accuracy and adaptability in structure state identification. The method first constructs a module template through the structural similarity between modules, combines a dynamic continuous feature extraction mechanism, and jointly models and classifies the structural variation and distributed abnormality in different test samples. Second, the multi-module joint feature and abnormal hotspot area model construction strategy is introduced to improve the perception ability of complex structures such as cross connection and fuzzy boundary. Finally, through the global consistency constraint and substructure feature alignment mechanism, the stability of the system in diversified scenarios is improved. Experimental results show that the method has significant advantages in abnormal identification accuracy and robustness compared with the comparative method without introducing the technical features in multiple test modules, and is suitable for state monitoring and abnormal warning tasks of complex structure systems.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of structural health detection of industrial systems, and particularly relates to a module connection automatic detection method and system based on template learning and abnormality positioning. BACKGROUND

[0002] In recent years, with the rapid development of industrial automation, electronic assembly, new energy vehicles, electrical control equipment and other industries, modular structures are widely used in product design and production manufacturing, especially in control boards, wire harness plug-in terminals, relay modules and other components. The module usually contains multiple electrical connection lines inside. As the basis for ensuring signal transmission, function control and power supply reliability, the correctness of the module connection directly relates to the operation stability and product quality of the entire system.

[0003] To ensure the correct connection of the module connection, manual inspection, fixed test tooling or preset conduction loop are generally used for point-by-point detection in the current manufacturing process. Such traditional methods have the following disadvantages: first, they rely on manual operation, which is labor-intensive and difficult to ensure consistency and accuracy; second, the detection efficiency is low, especially when dealing with module structures with a large number of connection terminals, dense wiring and complex combination methods, the detection period is long and the rework process is complicated; third, they cannot realize adaptive learning and intelligent judgment of the connection structure, making it difficult to meet the needs of flexible production and rapid switching of module varieties.

[0004] In addition, in actual applications, if the module connection has problems such as misconnection, missing connection, virtual connection or abnormal contact resistance, it may cause electrical signal interruption, control logic disorder and even safety hazards, seriously affecting product quality and customer experience. Therefore, how to improve the accuracy of connection detection, realize the rapid learning of standard templates of connection structure, and realize the real-time positioning and visual feedback of abnormal points in the detection process, has become a key problem that needs to be solved in the current technical field.

[0005] However, there is still a lack of an efficient detection mechanism that integrates "template learning-real-time detection-abnormality determination-structure positioning-interface prompt" in the prior art, and there is also a lack of a detection system scheme with structure universality, scalability and automatic adaptation capability. Therefore, it is of great practical significance and wide application prospect to develop a module connection automatic detection method and system based on template learning and abnormality positioning capability. SUMMARY

[0006] The present application aims at the above defects, and provides a module connection automatic detection method and system based on template learning and abnormal positioning, which aims at the problems of low detection efficiency, poor determination accuracy, and inability to dynamically adapt to multiple types of modules in the existing connection detection method, and proposes a method of constructing a standard module template and comparing it with a to-be-detected module in real time to realize rapid determination and accurate positioning of module connection. The method comprises the following steps: standard module feature data acquisition and template establishment, template library management, to-be-detected module data acquisition, feature comparison analysis, abnormal connection positioning, and visual prompt. Meanwhile, the system comprises a template import unit, a standard module acquisition unit, a template storage unit, a real-time acquisition unit, a comparison analysis unit, and an abnormal positioning unit, which are clear in division of labor and operate in cooperation, and can effectively support the realization of the automatic detection process.

[0007] The present application provides the following technical solutions: according to a first aspect of the present application, a module connection automatic detection method based on template learning and abnormal positioning is provided, comprising the following steps:

[0008] Starting the detection system and importing the module connection detection template;

[0009] Connecting a standard module with correct connection to the detection platform, acquiring electrical feature data of all connections in the standard module, and generating a corresponding connection reference model;

[0010] Storing the connection reference model in the template library for subsequent comparison and analysis;

[0011] Connecting a to-be-detected module to the detection system, acquiring electrical feature data of the connections of the to-be-detected module, and performing matching analysis with the connection reference model;

[0012] According to the matching analysis result, outputting connection correctness determination information, and if an abnormality is found, the system identifies the abnormal connection position.

[0013] In some embodiments, the step of connecting a standard module with correct connection to the detection platform, acquiring electrical feature data of all connections in the standard module, and generating a corresponding connection reference model comprises:

[0014] Acquiring multi-dimensional feature data of the connection, including on-resistance, signal delay, contact stability, and port mapping relationship;

[0015] Filtering and normalizing the acquired data to reduce the influence of test environment noise on the reference model;

[0016] Establishing a connection determination range based on an adaptive threshold algorithm, so that the judgment condition can be automatically adjusted according to different module types during subsequent detection.

[0017] In some embodiments, the step of connecting the to-be-tested module to the detection platform, controlling the detection system to collect the electrical characteristic data of the to-be-tested module, and performing matching analysis with the wiring reference model comprises:

[0018] connecting the to-be-tested module to the detection platform and starting the detection program;

[0019] collecting real-time electrical characteristic data of all the wirings of the to-be-tested module;

[0020] performing point-by-point comparison between the real-time data and the wiring reference model in the corresponding template library;

[0021] when an abnormality is detected, locating the physical interface position corresponding to the abnormal wiring and marking the position in real time on the detection interface;

[0022] In some embodiments, the step of locating the physical interface position corresponding to the abnormal wiring comprises:

[0023] based on the port mapping relationship and the test channel number, reverse searching the physical position of the abnormal wiring in the module structure;

[0024] visually displaying the abnormal point position in the detection interface in the form of highlighting, flashing or text prompts;

[0025] supporting storage and export of the history record of the abnormal point position for subsequent fault analysis and module repair;

[0026] supporting parallel positioning of multiple abnormalities during the detection process and outputting according to the severity of the abnormalities.

[0027] In some embodiments, the step of performing point-by-point comparison between the real-time data and the wiring reference model in the corresponding template library comprises:

[0028] calling the wiring reference model matching the to-be-tested module model from the template library and loading the multi-dimensional characteristic value range and port mapping information contained therein;

[0029] performing feature matching identification on each wiring in the to-be-tested module to obtain the corresponding real-time measurement data item;

[0030] according to the port mapping relationship of the wiring stored in the template, matching the reference data of the current wiring and the same port mapping wiring in the reference model one by one;

[0031] performing comparison and judgment on the real-time measurement data and the upper and lower threshold values of the corresponding characteristic values in the reference model, and if any characteristic value exceeds the preset threshold range, determining that the wiring is abnormal;

[0032] recording the comparison result in the detection system in the form of Boolean identifier, deviation value or grade score for subsequent abnormal marking, sorting or visual output.

[0033] In some embodiments, the step of reverse-checking the physical location of the abnormal connection in the module structure based on the port mapping relationship and the test channel number includes:

[0034] Obtain the standard port mapping relationship corresponding to the abnormal connection from the template library, including the input port number and the output port number;

[0035] Based on the actual test wiring diagram or preset port structure mapping table of the module under test, associate the test channel number assigned during the sampling process with the physical interface location in the module structure;

[0036] By comparing the mapping relationship between the test channel and the standard port, the physical port location information of the module corresponding to the input and output ends of the abnormal connection is determined, including one or more of the following: module housing number, port coordinates, row and column position or position label;

[0037] The physical location results obtained from the reverse lookup are mapped to a graphical module structure model for subsequent visual error prompts on the interface.

[0038] In some embodiments, feature matching and identification are performed on each connection in the module under test, and the corresponding real-time measurement data items obtained include one or more of the following: on-resistance, signal delay, and contact stability parameters.

[0039] According to a second aspect of this application, an automatic module connection detection system based on template learning and anomaly localization includes a template import unit, a standard module acquisition unit, a template storage unit, a real-time acquisition unit, and a comparison analysis and anomaly localization unit.

[0040] The template import unit is used to import the module connection detection template when the detection system is started;

[0041] The standard module acquisition unit is used to acquire electrical characteristic data of all connections in the standard module and generate a corresponding connection reference model when a standard module with known correct wiring is connected to the testing platform.

[0042] The template storage unit is used to store the connection benchmark model in the template library for subsequent comparative analysis;

[0043] The real-time acquisition unit is used to acquire electrical characteristic data of each connection of the module under test in real time after the module under test is connected.

[0044] The comparison analysis and anomaly location unit is used to perform matching analysis with the connection reference model after the module under test is connected; and output connection correctness judgment information based on the matching analysis results. If an anomaly is found, the system identifies the location of the abnormal connection.

[0045] According to a third aspect of the present application, an electronic device is provided, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the module connection automatic detection method based on template learning and abnormal positioning as described above when executing the computer program.

[0046] According to a fourth aspect of the present application, a computer-readable storage medium is provided, storing instructions which can implement the module connection automatic detection method based on template learning and abnormal positioning as described above when executed.

[0047] The present application can establish adaptive detection templates for modules of different structures by introducing a template learning mechanism, thereby improving the compatibility of the system for multiple types of modules. The use of multi-dimensional electrical feature comparison and adaptive threshold judgment improves the accuracy and robustness of the detection. The introduction of abnormal connection positioning and structure mapping mechanisms enables accurate positioning of abnormal connections and real-time interface prompts, significantly reducing the time for manual troubleshooting. The overall scheme has good scalability, high automation, and is suitable for module connection quality online monitoring and intelligent judgment in large-scale manufacturing scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0048] The present application will be described in more detail below based on embodiments and with reference to the accompanying drawings. Among them:

[0049] Figure 1 The structure block diagram of the module connection automatic detection system 100 based on template learning and abnormal positioning provided by the embodiments of the present application is shown;

[0050] Figure 2 The flowchart of the module connection automatic detection method 200 based on template learning and abnormal positioning provided by the present application is shown.

[0051] Figure 3 The flowchart of the method 300 of generating corresponding connection reference models in the embodiments of the present application is shown.

[0052] Figure 4 The flowchart of a module connection automatic detection method 400 based on template learning and abnormal positioning in the embodiments of the present application is shown.

[0053] Figure 5 The flowchart of a method 500 of positioning the physical interface position corresponding to the abnormal connection in the embodiments of the present application is shown.

[0054] Figure 6 The flowchart of a method 600 of point-by-point comparison of real-time data with connection reference models in the corresponding template library in the embodiments of the present application is shown.

[0055] Figure 7A flow chart of the method 700 for inversely searching the physical position of an abnormal connection in a module structure based on a port mapping relationship and a test channel number in the embodiments of the present application is shown.

[0056] Figure 8 A comparative line graph showing the accuracy variation of the method of the present application and Comparative Examples 1-4 in ten different test modules is shown. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0058] As described in the technical solutions proposed in the present application, in order to solve the problems of low detection efficiency, poor structural adaptability and inaccurate positioning of abnormal connections in the existing module connection detection method, the present application further proposes a module connection automatic detection system based on template learning and abnormal positioning. The system combines multiple functional modules such as standard module feature learning, real-time feature acquisition, feature comparison analysis and abnormal structure positioning, and can realize rapid judgment of the internal connection state of different types of complex modules and visual feedback of fault points, thereby effectively improving the intelligentization and automation level of module detection.

[0059] In order to better implement the detection method, the detection system of the present application adopts modular design, which not only supports the rapid import and calling of detection templates, but also has the ability of unified management of module structure data and graphical output of abnormal positioning. The following will describe in detail the various constituent units of the module connection automatic detection system of the present application with reference to the drawings, so as to make the structure and working process of the system of the present application more clearly understood by those skilled in the art.

[0060] Figure 1 A structural block diagram of the module connection automatic detection system 100 based on template learning and abnormal positioning provided by the embodiments of the present application is shown, which includes a template import unit 101, a standard module acquisition unit 102, a template storage unit 103, a real-time acquisition unit 104, a comparison analysis and abnormal positioning unit 105. The various functional units cooperate with each other to form a closed-loop detection system that can complete template learning, real-time detection and abnormal positioning feedback.

[0061] The template importing unit 101 is configured to call and import a detection template corresponding to a structure of a module to be detected during a system startup stage. The template contains port mapping information, connection logic relationship, and an electric characteristic parameter range that can be learned of a standard module structure. Through the template importing, the system can quickly match a corresponding detection configuration according to a detection task type, and realize quick switching and adaptation of different module categories.

[0062] The standard module collecting unit 102 is configured to automatically collect key electric characteristic data of each connection after a standard module is connected to the detection platform. The data includes, but is not limited to, conduction resistance, signal delay, contact stability and other multi-dimensional characteristic parameters. The system constructs a connection reference model of the standard module according to the collection result, and provides a template basis for subsequent comparison and analysis.

[0063] The template storage unit 103 is configured to store the connection reference model generated by the standard module collecting unit into a template library. The template library can support concurrent management of multiple module structure models, and has functions of model identification, calling interface, version control and the like, so as to ensure that historical templates can be quickly and accurately called and updated and replaced during the detection process.

[0064] The real-time collecting unit 104 is configured to collect electric characteristic data of all connections of the module to be detected in real time after the module is connected. The collection manner is consistent with that of the standard module collection, so as to ensure consistency and comparability of the data. The collected real-time data will be used as detection input, and compared and analyzed with the template data.

[0065] The comparison and analysis and abnormal positioning unit 105 is configured to compare and analyze the test data acquired by the real-time collecting unit with the connection reference model extracted from the template storage unit. If the electric characteristic data of a connection is out of a normal range set by the template, the system automatically determines that the connection is abnormal, and determines a physical position of the abnormal connection according to a port mapping relationship and a test channel number, and finally performs abnormal prompt and graphical highlighting display in a detection interface, so as to facilitate an operator to locate a fault point.

[0066] Figure 2 A flowchart of a module connection automatic detection method 200 based on template learning and abnormal positioning provided by the present application is shown. The method can be executed by the detection system, and is suitable for connection correctness detection of electronic modules, control modules or wire harness assemblies of various structure types, and is particularly suitable for module products with complex connection structures.

[0067] S210: Start the detection system and import a module connection detection template.

[0068] In this step, the operator starts the detection system, the system loads the user permission information and initializes the detection environment. The operator can select the detection template corresponding to the current module type in the interface, which contains the module structure description, port mapping table and historical learning connection reference data, providing a basis for subsequent standard module learning or detection process.

[0069] S220: Connect the standard module with known correct connection to the detection platform, collect the electrical characteristic data of all connections in the standard module, and generate the corresponding connection reference model.

[0070] In this step, the operator connects a standard module with known structure and completely correct connection to the detection system. The system measures the multi-dimensional electrical characteristic parameters of each connection in the module, including on-resistance, signal delay, contact stability, etc. Through certain data processing methods (such as filtering, normalization), a stable and reliable connection reference model is generated, which is used to define the detection standard for this type of module.

[0071] S230: Store the connection reference model in the template library for subsequent comparative analysis.

[0072] The generated connection reference model will be saved to the system template library and bound with the module model or structure number. The template library supports various template management operations, including version control, overlay update, template export, etc., ensuring that subsequent detection tasks can directly call the corresponding template without the need to repeatedly collect standard module data.

[0073] S240: Connect the module to be tested to the detection system, collect the connection electrical characteristic data of the module to be tested, and perform matching analysis with the connection reference model.

[0074] When the operator connects the module to be tested, the system will collect the real-time electrical characteristic data of each connection in the module in the same way as the standard module. Then, the system compares the collected data with the corresponding connection reference model in the template library one by one to determine whether each connection is within the normal threshold range set by the template.

[0075] S250: According to the matching analysis result, output the connection correctness judgment information, if abnormal, the system identifies the abnormal connection position.

[0076] The system generates a detection report according to the comparison result and outputs the overall judgment result of the module connection. When a connection is identified as abnormal, the system will accurately locate the physical port or interface number connected by the abnormal connection in combination with the port mapping table and structure mapping data, and visually highlight the prompt in the detection interface to assist the operator in maintenance, retesting or traceability processing.

[0077] Figure 3A flow chart of a method 300 for generating a corresponding connection reference model in the embodiments of the present application is shown.

[0078] In step S310, multi-dimensional feature data of the connection is collected. For the ith connection , the following electrical features are collected: on-resistance , signal delay , contact stability , and port mapping relationship .

[0079] For the on-resistance , a four-wire measurement is used, with current excitation and voltage sampling. The measurement formula is as follows: ;

[0080] Wherein is the voltage measured for the ith connection, is the excitation current, with units of ohms (Ω).

[0081] For the signal delay , the impulse response method is used to measure the response delay of ports A and B through time difference: , which can be used to determine connection length abnormalities or capacitive interference. , are the response time of the ith connection at port B and the trigger time of port A, respectively, and the difference between the two is the response delay.

[0082] For the contact stability , it can be measured by voltage fluctuation amplitude, transient change rate, etc. For example, within a specified time window , the standard deviation of the voltage or resistance value measured continuously times is calculated:

[0083] ;

[0084] to determine whether the terminal has a virtual weld or poor contact. In this formula, x is the voltage or resistance value, as the independent variable, is the jth sample value of the independent variable of the ith connection. When x is the voltage, is ; when x is the resistance, is . is the average value of the corresponding independent variable over N statistical samples.

[0085] For the port mapping relationship , for each connection, define a unique mapping pair of its input port and output port : The mapping relationship is stored in the connection template. 、 are the i-th connection input port number and output port number, respectively.

[0086] In step 320, the collected data is filtered and normalized. Generally, moving average filtering or median filtering is used to remove random noise:

[0087] If moving average filtering is used (taking the on-resistance as an example):

[0088] ;

[0089] is the on-resistance value of the i-th connection after filtering and smoothing. is the original on-resistance value of the i-th connection at the j-th sampling time, with units of ohms (Ω). k is the length of the sliding window, and n is the current sampling time index.

[0090] If median filtering is used (taking the contact voltage as an example): is the contact voltage value of the i-th connection after filtering and smoothing, and median() is the median function, is the contact voltage value of the i-th connection at the j-th sampling time, with units of volts (V).

[0091] Normalization is to normalize all feature quantities to the [0, 1] interval, which is convenient for subsequent modeling. The normalization calculation formula for signal delay is as follows:

[0092] ;

[0093] where 、 are the maximum and minimum values of the signal delay in all connections, respectively. is the signal transmission delay time of the i-th connection, with units of seconds (s), representing the time required for the electrical signal to transmit from one end to the other end.

[0094] The normalized multi-dimensional features can form a unified feature vector for comparison:

[0095] .

[0096] is the on-resistance normalized value of the i-th connection, is the signal transmission delay time normalized value of the i-th connection, and is the contact stability feature quantity of the i-th connection.

[0097] In step S330, the system obtains the statistical feature range of each type of connection through historical learning or standard module training sets, and then sets an "adaptive judgment interval" based on its distribution characteristics to determine whether the test value falls within this interval. Specifically, this includes the following steps:

[0098] Obtain the feature sample set of the same type of connection in the standard module. ;

[0099] Calculate the mean μ and standard deviation σ;

[0100] Define upper and lower thresholds: ;

[0101] Judgment conditions: .

[0102] Where η is the sensitivity coefficient, which is usually taken as 2 to 3 to cover 95 to 99% of the sample range.

[0103] This step can improve detection accuracy and adaptability.

[0104] Figure 4 A flowchart of an automatic module wiring detection method 400 based on template learning and anomaly localization, according to an embodiment of this application, is shown. This method, during the detection phase, involves connecting the module under test to the detection platform, collecting electrical characteristic data of its wiring in real time, and comparing and analyzing this data with a pre-learned wiring benchmark model to automatically determine the correctness of the wiring and provide anomaly localization prompts. The method includes the following steps:

[0105] S410: Connect the module under test to the testing platform and start the testing program. The operator connects the interface of the module under test to the system's testing channel through the interface or automatic identification, and triggers the testing system to enter sampling mode.

[0106] S420: Acquire real-time electrical characteristic data of all connections in the module under test. The detection system collects data from each connection in the module under test according to a preset sampling strategy. The characteristic data includes, but is not limited to, parameters such as on-resistance, signal delay, and contact voltage stability. The data is then preprocessed and normalized to ensure consistency with the template model.

[0107] The S420 steps are as follows:

[0108]

[0109] S430: The system performs point-by-point comparison between real-time data and the connection baseline model in the corresponding template library. The system extracts the standard connection model that matches the current module structure from the template storage unit, and compares the collected feature vectors with the standard values ​​of each connection in the template to identify whether there are any electrical characteristic deviations.

[0110] S440: When an anomaly is detected, the physical interface position corresponding to the abnormal connection is located, and a prompt is marked in real time in the detection interface. If the characteristic parameter of a connection exceeds the threshold interval set in the template, the system determines that the connection has an anomaly, and according to the mapping relationship between the port number and the test channel, the physical position is accurately searched. At the same time, the visual identification is performed through color highlighting, flashing icons or text prompts in the interface, so as to facilitate the operator to quickly judge the fault connection and repair or retest.

[0111] The steps S430 and S440 are as follows:

[0112] .

[0113] Figure 5 A flow chart of a method 500 for locating the physical interface position corresponding to the abnormal connection in the embodiments of the application is shown. The method is used to accurately locate the specific physical position of the abnormal connection in the module structure based on the port mapping relationship and channel number information established in the detection system after the abnormal connection is found in the connection comparison detection, and to prompt in a visual way in the detection interface, so as to improve the efficiency of abnormal handling and the convenience of user operation. The method includes the following steps:

[0114] In the step S510, the system automatically analyzes the module interface and its structure position corresponding to the channel according to the mapping table between the module port and the test channel in combination with the channel number marked as abnormal in the detection process based on the port mapping relationship and the test channel number, and searches the physical position of the abnormal connection in the module structure, so as to realize the reverse positioning of the abnormal point. Specifically as follows:

[0115]

[0116] In the step S520, in order to facilitate the operator to quickly identify the abnormal position, the system performs graphical identification on the abnormal connection in the detection main interface, and the common forms include red highlight frame, flashing symbol mark or matching abnormal description text, so that the detection interface has instant interactivity and intuitiveness. Specifically as follows:

[0117]

[0118] In the step S530, the system stores the information such as the detected abnormal connection number, detection time, characteristic deviation value and positioning result in a structured way, and supports exporting in the form of table, report or image, so as to facilitate the operation and maintenance personnel to analyze, attribute and maintain and manage the abnormal module afterwards.

[0119] In step S540, for the case that multiple abnormal connections exist in the same module, the system can simultaneously locate all abnormal points, and generate an abnormal priority ranking list in combination with the deviation amplitude or multi-dimensional feature deviation degree of each abnormal item, to guide the operator to prioritize the processing of critical failures. The specific process is as follows:

[0120]

[0121] Figure 6 A flow chart of the method 600 for point-by-point comparison between real-time data and connection reference models in the corresponding template library in the embodiments of the present application is shown. The method 600 is used to match and compare each item between the real-time feature data of the module to be tested and the standard model pre-stored in the template library in the module connection detection process, so as to determine whether the state of each connection is qualified, thereby providing basic support for subsequent abnormal positioning and detection result output.

[0122] In step S610, after the detection system completes the connection of the module to be tested, the module number, model number or structural feature identifier is first read, and the connection reference model matched with the module is searched in the template library. The feature matching identification of each connection in the module to be tested is performed, and one or more of the obtained corresponding real-time measurement data items include the turn-on resistance, signal delay, contact stability parameter. The reference model includes: the standard port mapping relationship of each connection, the corresponding turn-on resistance, signal delay, contact stability and other multi-dimensional reference feature values and their preset threshold range (such as μ±ησ, IQR boundary, etc.).

[0123] In step S620, the system collects the electrical features of all connections in the module to be tested through a multi-channel sampling device, including but not limited to the turn-on resistance value, excitation response delay time, port-to-port fluctuation stability, etc. Each connection generates a set of real-time feature vectors , for example, .

[0124] In step S630, the system searches for the to-be-tested connections having the same port mapping according to the port pairs (such as P1-P7, P3-P10, etc.) corresponding to each standard connection in the template model in the real-time collected data. If there is a one-to-one mapping, the association between the real-time connection and the template connection is established.

[0125] In step S640, for each pair of connections having established mapping relationship, the system compares the real-time feature values with the reference feature value interval in the template one by one, and judges whether it is out of the adaptively set range. If any feature is out of the range, the connection is marked as abnormal. The judgment method can be based on hard threshold judgment (greater than / less than), Z-score deviation value, robust outlier discrimination, etc.

[0126] In step S650, the system records the comparison results of all connections in a structured manner, including connection number, anomaly type, actual value, and standard value range. For anomalies, parameters such as severity score and offset distance from the template center can be attached to facilitate subsequent sorting, visualization prompts, and export of detection reports.

[0127] In a preferred embodiment, method 600 may further include a step of dynamically calculating the percentage of anomalies and the global error trend during the comparison process, which is used to further intelligently adjust the judgment strategy or prompt the module structure for anomalies.

[0128] In other embodiments, the comparison process described above in method 600 can be executed on a backend server or edge computing node, separating remote access from on-site inspection via a template database, further improving inspection efficiency and system versatility. The comparison results can be synchronized in real time to a web interface, touchscreen terminal, or industrial control system for visualization and rework decision-making.

[0129] Figure 7 This document illustrates a flowchart of a method 700 for reverse-engineering the physical location of an abnormal connection within a module structure based on port mapping relationships and test channel numbers, as described in an embodiment of this application. Method 700, after the detection system identifies an abnormal module connection, uses preset standard port mapping information and test channel allocation records from a template library to deduce and locate the specific physical interface position of the abnormal connection within the module structure. This enables reverse tracing and precise mapping of the connection fault location, providing spatial location information support for subsequent visualization and fault repair operations.

[0130] In step S710, after detecting a connection anomaly, the system first reads the port mapping information corresponding to the abnormal connection from the loaded connection template, including the input port number (e.g., P3) and the output port number (e.g., P10). This mapping relationship is usually generated by the standard module during the learning phase and has uniqueness and locatability.

[0131] In step S720, after the module is connected to the testing platform, the testing system binds the hardware test channel number (such as CH01, CH02, etc.) used during actual sampling to the corresponding physical interface port on the module. This binding information can be pre-established through a structure mapping table (such as a structure JSON file, tooling interface number diagram, or BOM mapping relationship) or automatically loaded after identifying the module model.

[0132] In step S730, the system determines which specific physical location on the module structure each port (P3 and P10) is located at by comparing the standard port mapping relationship with the current test channel allocation. For example, port P3 is located at the lower left corner and is numbered "J1-04", and port P10 is located at the upper right side and is numbered "J2-08". In some embodiments, the physical positioning result can also be provided in the form of port coordinates (such as X, Y positions) or location labels (such as "Side A, 3rd row, 2nd column").

[0133] In step S740, the system maps the physical information of the abnormal point obtained by the reverse search to the module structure model diagram or the interactive interface in the detection platform interface. The abnormal point can be visually displayed in the form of red highlighting, flashing identification, or a suspended label of the port number, so that the operator can directly locate the connection position of the abnormal cable and quickly implement repair.

[0134] In some other embodiments, the method 700 further includes a step of automatically synchronizing the reverse positioning result to the detection report and exporting it in the form of a structural diagram. In combination with a module CAD drawing or a 3D visualization tool, the system can implement functions such as abnormal distribution heat map and interface abnormal clustering statistics at the structure level, which are used for subsequent batch quality analysis and process tracing.

[0135] Figure 8 A comparison line graph showing the accuracy variation of the inventive method and Comparative Examples 1-4 in ten different test modules is shown, which is used to show the performance difference of each method in terms of continuous structure recognition and abnormal classification ability. The abscissa is the serial number of the test module to be detected, and the ordinate is the accuracy of abnormal positioning. A total of 10 test modules were statistically analyzed for continuous structure recognition. Figure 8 The yellow line in the graph is the accuracy of abnormal connection positioning of the test module using the module connection automatic detection method 200 based on template learning and abnormal positioning provided in the present application in Comparative Example 1. In Comparative Example 1, the corresponding connection reference model generated by method 300 was used, and method 400 in the present application was also used in the connection reference model matching analysis process when the test module was connected to the detection platform. In the process of positioning the physical interface position of the abnormal connection, the method 700 provided in the present application was also used in the step of reverse searching the physical position of the abnormal connection in the module structure based on the port mapping relationship and the test channel number. In the step of point-by-point comparison of real-time data and the connection reference model in the corresponding template library in method 400 of Comparative Example 1, the method 600 provided in the present application was also used.

[0136] Figure 8The orange fold line in the figure is Comparative Example 2, in which the method 300 in the embodiments of the present application is not used to connect the standard module with known correct connection to the detection platform, collect the electrical characteristic data of all connections in the standard module, that is, the method based on module template learning and continuous reference model construction, so that the overall stability decreases in the test samples with complex structure or discontinuous abnormal jump characteristics. This method uses static threshold rules and local statistical thresholds to classify and judge the initial connection state of the module structure, and does not establish a coherent structure mapping and dynamic abnormal feature trajectory association.

[0137] From Figure 8 The trend of the orange fold line in the figure can be seen. In most test samples, this method still has a certain degree of abnormal structure recognition ability, but its accuracy is generally lower than that of the overall method proposed in the present application, especially in samples 2, 4 and 8, which fluctuate sharply to about 0.79, 0.77 and 0.78 respectively, indicating a lack of adaptability analysis capability for the continuous mode between modules. When facing test samples with complex jump paths between samples, this method is difficult to build an abnormal perception view from the overall structure, which limits its overall recognition effect.

[0138] In addition, from the overall fold line trend, it can be observed that the method of Comparative Example 2 shows a certain stability under each module test, but fails to effectively alleviate the influence of cross-module structure misplacement or local noise interference, resulting in limited upper limit of accuracy under diverse samples, indicating that its model lacks universal adaptability to structural mutations and lacks a cross-module structural context modeling mechanism.

[0139] The red fold line is Comparative Example 3, in which the method 400 in the embodiments of the present application is not used, so when performing matching analysis of the connection relationship of the to-be-tested module, only static matching based on the preset connection position relationship is performed, without combining the bidirectional mapping mechanism based on module port number and connection path proposed in the present application. Therefore, in actual scenarios, the non-fixedness of module connection, reverse plugging of ports or multi-port fuzzy mapping problems are not fully considered, which affects the recognition ability of abnormal connections.

[0140] As Figure 8 shown, the accuracy of the red fold line in each test sample is lower than that of Comparative Example 1 (yellow fold line) using the complete method, especially in the 4th and 8th test modules, the accuracy decreases to about 0.69 and 0.70 respectively, which has a large deviation, verifying that when the module port mapping relationship analysis method (i.e., matching analysis with the connection reference model) involved in the method 400 is not used, the abnormal connection positioning is easily affected by interference factors, resulting in performance degradation.

[0141] Figure 8The blue fold line in the figure is the comparative example 4. In the comparative example 4, the method 700 provided in the embodiment of the present application is not used, that is, the reverse lookup method based on the test channel number and the module structure mapping is not used in the process of positioning the physical interface position corresponding to the abnormal connection, but only relies on the static information recorded in the connection table. In this way, the dynamic change of the module position in the actual connection deployment and the non-one-to-one mapping relationship between the test number and the physical structure are ignored.

[0142] The accuracy of the blue fold line is slightly higher than that of the comparative example 5 or is the same as that of the comparative example 3 in some modules, but is still significantly lower than that of the comparative example 1 corresponding to the method of the present application. Especially in the second, fourth and eighth modules, the accuracy is not higher than 0.74, indicating that the system has certain deficiencies in the abnormal positioning ability of the internal connection of the complex module under the condition of lacking the structure reverse lookup mechanism.

[0143] Figure 8 The pink fold line in the figure is the comparative example 5. In the comparative example 5, the point-by-point comparison method of the real-time data and the template library connection reference model involved in the method 600 in the embodiment of the present application is not used, only the global error or the historical average is used for abnormal judgment, and the single connection or local abnormal point cannot be accurately analyzed, so the recognition ability of the method for the small connection abnormality is limited.

[0144] From Figure 8 It can be observed from the pink fold line in the figure that the accuracy in most test modules is between 0.68 and 0.73, and in the fourth and seventh test modules, it is particularly low, which is 0.66 and 0.65 respectively, indicating that the method is relatively weak in dealing with noise disturbance and atypical connection abnormality.

[0145] Comprehensive Figure 8 It can be seen from the five fold lines in the figure that the method (comparative example 1, yellow fold line) provided in the present application maintains an abnormal connection positioning accuracy of 0.95 or more in all 10 test modules, and the average accuracy reaches 0.97 or more, which is much higher than that of other comparative examples. The method of the comparative example 1 comprehensively introduces the method 300-the method 700, which not only ensures the accurate generation of the connection reference model, but also improves the robustness of the connection relationship matching, especially in the abnormal positioning stage through real-time comparison and structure reverse lookup mechanism, which significantly improves the accuracy and stability of the system.

[0146] In the embodiment of the present application, the module connection automatic detection method can not only be realized through a hardware detection platform and a matching software system, but also can be deployed on an electronic device with computing and storage functions, so as to further expand the applicability and deployment flexibility thereof.

[0147] The embodiment of the present application further provides an electronic device, which comprises a memory and a processor, the memory stores a computer program, and the processor implements the module connection automatic detection method based on template learning and abnormal positioning as described in the embodiment of the present application when executing the computer program. Specifically, the processor can be a general central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, or a combination of the above processing units. The memory can include a volatile memory (such as a random access memory RAM) and / or a non-volatile memory (such as a read only memory ROM, a flash memory, a hard disk drive HDD or a solid state drive SSD, etc.), used to store program codes, template data, connection reference models and real-time feature data collected during detection. During the running process of the electronic device, by loading the computer program, the steps of template import, standard module learning, template storage, real-time data collection, feature comparison analysis and abnormal positioning can be sequentially completed, so that the rapid and accurate detection of the connection state of the module to be tested can be realized, and the specific position and determination information of the abnormal connection can be output.

[0148] In addition, the embodiment of the present application further provides a computer readable storage medium, which stores computer instructions, and the computer instructions implement the module connection automatic detection method based on template learning and abnormal positioning as described above when executed by a processor. The computer readable storage medium can be a magnetic storage medium (such as a magnetic disk or a magnetic tape), an optical storage medium (such as a compact disc CD or a digital versatile disc DVD), a semiconductor storage medium (such as a flash disk or a solid state disk), or any combination of the above media. When implementing the method of the present application, the instructions can be loaded into the processor of the electronic device for running, to realize the functions of template data calling, real-time electrical feature data acquisition, normalization processing, multi-dimensional feature comparison, abnormal determination and visual marking, to ensure that the detection system can maintain high efficiency and stable detection performance in complex and variable connection environment, and to provide reliable data support for module quality control and fault repair.

[0149] Although the present application has been described with reference to the preferred embodiments, various modifications and changes can be made thereto without departing from the scope of the application. In particular, the technical features mentioned in the various embodiments can be combined in any manner, as long as there is no technical or conceptual contradiction. The present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for automatic detection of module interconnections based on template learning and anomaly localization, characterized in that, The method comprises the following steps: starting a detection system and importing a module connection detection template; connecting a standard module with correct connections to a detection platform, collecting electrical characteristic data of all connections in the standard module, and generating a corresponding connection reference model; storing the connection reference model in a template library for subsequent comparative analysis; connecting a module to be tested to the detection system, collecting electrical characteristic data of connections of the module to be tested, and performing matching analysis with the connection reference model; when an abnormality is detected, locating a physical interface position corresponding to the abnormal connection, and marking a prompt in real time on a detection interface; the step of locating the physical interface position corresponding to the abnormal connection comprises: based on a port mapping relationship and a test channel number, inversely searching a physical position of the abnormal connection in a module structure; visually displaying the abnormal point position in a highlighted, flashing or text prompt manner on the detection interface; supporting storage and export of a history record of the abnormal point position for subsequent fault analysis and module repair; supporting parallel positioning of multiple abnormalities during detection, and outputting according to an abnormality severity ranking; the step of based on the port mapping relationship and the test channel number, inversely searching the physical position of the abnormal connection in the module structure comprises: obtaining a standard port mapping relationship corresponding to the abnormal connection from the template library, including an input port number and an output port number; according to an actual test wiring diagram or a preset port structure mapping table of the module to be tested, associating a test channel number allocated in a sampling process with a physical interface position in the module structure; comparing a mapping relationship between the test channel and the standard port, determining module physical port position information corresponding to an input end and an output end of the abnormal connection, including one or more of a module shell number, a port coordinate, a row-column position or a position label; mapping the inversely searched physical position result to a graphical module structure model for subsequent visual abnormality prompting on the interface.

2. The method of claim 1, wherein, The step of connecting the standard module with correct connections to the detection platform, collecting electrical characteristic data of all connections in the standard module, and generating a corresponding connection reference model comprises: collecting multi-dimensional characteristic data of the connections, including a conduction resistance, a signal delay, a contact stability and a port mapping relationship; performing filtering and normalization processing on the collected data to reduce an influence of test environment noise on the reference model; based on a self-adaptive threshold algorithm, establishing a connection judgment range, so that the judgment condition can be automatically adjusted according to different module types during subsequent detection.

3. The method of claim 1, wherein, The step of connecting the module to be tested to the detection platform, controlling the detection system to collect electrical characteristic data of connections of the module to be tested, and performing matching analysis with the connection reference model comprises: connecting the module to be tested to the detection platform, and starting a detection program; obtaining real-time electrical characteristic data of all connections of the module to be tested; performing point-by-point comparison of the real-time data with the connection reference model in the corresponding template library; when an abnormality is detected, locating a physical interface position corresponding to the abnormal connection, and marking a prompt in real time on a detection interface.

4. The method of claim 3, wherein, The step of performing point-by-point comparison of the real-time data with the connection reference model in the corresponding template library comprises: Call the connection reference model matching the model of the module to be tested from the template library, and load the multi-dimensional characteristic value range and port mapping information it contains; Perform feature matching identification on each connection in the module to be tested to obtain corresponding real-time measurement data items; According to the port mapping relationship of the connection stored in the template, match the reference data of the connection with the same port mapping in the reference model one by one; Compare the real-time measurement data with the upper and lower threshold values of the corresponding characteristic values in the reference model, and if any characteristic value exceeds the preset threshold range, determine that the connection is abnormal; Record the comparison results in the detection system in the form of Boolean identification, deviation value or grade score for subsequent abnormal marking, sorting or visual output.

5. The method of claim 4, wherein, The corresponding real-time measurement data items obtained by performing feature matching identification on each connection in the module to be tested include one or more of the on-resistance, signal delay, and contact stability parameters.

6. A module interconnection automatic detection system based on template learning and abnormality positioning, characterized in that, It comprises a template import unit, a standard module acquisition unit, a template storage unit, a real-time acquisition unit, a comparison analysis and abnormal positioning unit; The template import unit is used to import the module connection detection template when starting the detection system; The standard module acquisition unit is used to acquire the electrical characteristic data of all connections in the standard module when the standard module with correct connections is connected to the detection platform, and generate a corresponding connection reference model; The template storage unit is used to store the connection reference model in the template library for subsequent comparison and analysis; The real-time acquisition unit is used to acquire the electrical characteristic data of each connection of the module to be tested in real time after the module to be tested is connected; The comparison analysis and abnormal positioning unit is used to perform matching analysis with the connection reference model after the module to be tested is connected; when an abnormality is detected, the physical interface position corresponding to the abnormal connection is located, and a real-time prompt is marked on the detection interface; The step of locating the physical interface position corresponding to the abnormal connection comprises: Based on the port mapping relationship and the test channel number, the physical position of the abnormal connection in the module structure is inversely searched; The abnormal point position is visually displayed on the detection interface in the form of highlighting, flashing or text prompt; Supporting historical record storage and export of the abnormal point position for subsequent fault analysis and module repair; Supporting multiple abnormal parallel positioning during detection, and outputting according to the severity of the abnormality; The step of locating the physical interface position corresponding to the abnormal connection based on the port mapping relationship and the test channel number comprises: Obtain the standard port mapping relationship corresponding to the abnormal connection from the template library, including the input port number and the output port number; According to the actual test wiring diagram or the preset port structure mapping table of the module to be tested, associate the test channel number allocated in the sampling process with the physical interface position in the module structure; Compare the mapping relationship between the test channel and the standard port to determine the module physical port position information corresponding to the input end and the output end of the abnormal connection, including one or more of the module shell number, the port coordinate, the row and column position, or the position label. The physical position result obtained by the reverse search is mapped into a graphical module structure model, so as to provide visual abnormality prompt in the interface subsequently. 7.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor implements the template learning and abnormality positioning-based module connection automatic detection method according to any one of claims 1-5 when executing the computer program.

8. A computer readable storage medium storing instructions, wherein, The instructions can implement the template learning and abnormality positioning-based module connection automatic detection method according to any one of claims 1-5 when executed.

Images