Automatic testing methods, systems, equipment, and storage media for automatic switching devices.

Abstract

This invention discloses an automatic testing method, system, device, and storage medium for automatic switching devices (AS / RS). The method includes: preprocessing and semantically parsing the original setting file of the target AS / RS to obtain a structured setting semantic object; using a knowledge graph of AS / RS multi-mode protection testing to match operating modes and select paths to obtain target test item paths and construct target test templates; sending the target test template to the target test terminal to control the target test terminal to execute simulated test actions; receiving actual test execution data and performing logical verification and processing to obtain a target inspection report. This invention can generate test paths based on the actual AS / RS mode, avoiding test item omissions and logical errors caused by misunderstandings or oversights when manually arranging test sequences for each mode, thus improving test completeness, accuracy, and automation.

Description

Technical Field

[0001] This invention relates to the field of intelligent testing of automatic switching devices for backup power supplies, and more particularly to an automatic testing method, system, equipment, and storage medium for automatic switching devices for backup power supplies. Background Technology

[0002] Automatic transfer switch (ATS) devices are crucial secondary devices in power systems used to automatically switch to backup power after an abnormal loss of the main power source. Their operational logic and response timing directly impact the reliability of power restoration. With the increasing number of substation devices and shortened maintenance cycles, the periodic inspection of ATS devices needs to complete tests of various operating modes, control functions, and protection functions within a limited maintenance window. In particular, different ATS modes, such as incoming line inter-transfer, sectional ATS, remote ATS, bridge ATS, and main transformer ATS, have different operating conditions, interlocking conditions, control word configurations, and test sequences. Therefore, it is necessary to generate test tasks matching the corresponding operating modes based on the actual setting configuration of the target ATS device, and to manage the test process and results uniformly.

[0003] In the current inspection process of automatic transfer switch (ATS) devices, maintenance personnel typically obtain setting sheets from the dispatch management system, electronic archives, or paper documents. They manually read parameters such as ATS mode, control words, action settings, and action times, and interpret each control bit of the control word according to the device manual. Subsequently, based on the ATS device's operating mode and functional activation / deactivation status, maintenance personnel set simulated voltage, simulated current, switch status, action sequence, and expected criteria item by item in the test terminal to form a test template. The test terminal then performs normal state simulation, fault state triggering, action output monitoring, and contact status acquisition on the ATS device according to the test template. After the test is completed, maintenance personnel verify the action data and waveform data and manually compile an inspection report. Therefore, manually arranging test sequences for each mode is extremely tedious and prone to omissions or logical errors in the template due to misunderstandings or oversights. In addition, the entire process involves isolated data links, and test reports are mostly paper or independent electronic documents, unable to be linked with setting management, equipment status assessment, and other systems, failing to form a digital closed loop. Summary of the Invention

[0004] The main objective of this invention is to provide an automatic testing method, system, device, and storage medium for automatic switching devices. This invention aims to solve the technical problems in existing technologies, such as the digital fragmentation of the fixed value data processing process relying on manual intervention, the extreme difficulty in control word parsing and multi-mode logic generation, the isolation of business processes, and the lack of automated closed loops, which lead to incorrect test parameters, missing test items, and a disconnect between test data and inspection results.

[0005] To achieve the above objectives, the present invention provides an automatic testing method for standby automatic switching devices, the method comprising the following steps: The original setting file of the target standby automatic switching device is preprocessed, and the preprocessed original setting file is semantically parsed to obtain a structured setting semantic object. Based on the pre-built knowledge graph of backup self-starting multi-mode protection test, the structured fixed-value semantic object is matched with the operation mode and the path is selected to obtain the target test item path. The target test template is constructed based on the target test item path. The knowledge graph of backup self-starting multi-mode protection test takes the backup self-starting mode as the top-level classification node, and each top-level classification node corresponds to a logical subgraph of different backup self-starting modes. The target test template is sent to the target test terminal, so that the target test terminal performs a simulated test on the target standby automatic transfer device based on the target test template. The system receives the actual execution data returned by the target test terminal, performs logical verification and processing on the actual execution data, and obtains the target inspection report.

[0006] Optionally, the preprocessing of the original setting file of the target standby automatic switching device and the semantic parsing of the preprocessed original setting file to obtain a structured setting semantic object include: The original setpoint file of the target automatic switching device is subjected to image enhancement, distortion correction and binarization processing to obtain a standardized image; The standardized image is input into the layout segmentation network, which outputs a table region mask and a text description region mask. Text detection and character recognition are performed based on the table region mask and the text description region mask to obtain a fixed-value single text sequence; The fixed-value single text sequence is input into the entity recognition model to extract the entity triples in the fixed-value single text sequence; The control words in the entity triples are decoded based on a preset protection model library to obtain a control word semantic set; Logical conflict verification is performed on the entity triplet and the control word semantic set based on the preset relay protection coordination rules. If the verification passes, a structured fixed-value semantic object is generated by combining the entity triplet and the control word semantic set.

[0007] Optionally, the step of performing text detection and character recognition based on the table region mask and the text description region mask to obtain a fixed-value single text sequence includes: Based on the table region mask and the text description region mask, extract the table region image and the text description region image from the standardized image; Text detection is performed on the table area image and the text description area image using a text detection model to obtain a set of text line coordinates. The set of text line coordinates includes the text line coordinates of each cell in the table area and the text line coordinates in the text description area. The standardized image is cropped based on the set of text line coordinates to obtain text line sub-images, and the size of the text line sub-images is normalized to obtain a text line image sequence. The text line image sequence is input into a pre-built character recognition model, which outputs a set of text strings; Based on the set of text line coordinates, the set of text strings is sorted by position and the text is combined to obtain a fixed-value single text sequence; The character recognition model consists of a convolutional neural network feature extraction module, a bidirectional long short-term memory network module, and a connection-sequence classification module. The convolutional neural network feature extraction module is configured to perform local visual feature extraction on the text line image sequence to obtain a text line visual feature sequence. The bidirectional long short-term memory network module is configured to perform bidirectional context encoding on the visual feature sequence of the text line to obtain the context feature sequence of the text line. The connection time-series classification module is configured to perform character category prediction, duplicate character merging, and whitespace marker deletion on the text line context feature sequence, and output a set of text strings.

[0008] Optionally, the step of decoding the control words in the entity triples based on a preset protection model library to obtain a control word semantic set includes: Extract the target control word and device model from the entity triple; According to the device model, the corresponding target control word definition table is extracted from the preset protection model library; The protection function corresponding to each bit of the target control word is determined according to the target control word definition table; The function enable / disable status is determined based on the target value of the target control word at each word position and the protection function corresponding to each word position. The function activation / deactivation status corresponding to all characters is combined to obtain a control word semantic set. The control word semantic set consists of multiple control word semantic tags, and the control word semantic tags consist of protection functions and function activation / deactivation status.

[0009] Optionally, the process of matching the operating mode and selecting the path of the structured fixed-value semantic object based on the pre-built knowledge graph of backup self-starting multi-mode protection test to obtain the target test item path, and constructing a target test template based on the target test item path, includes: Extract the self-starting mode setting, control word semantic set, setting item set, and logical condition set from the structured setting semantic object; Based on the set value of the backup automatic transfer mode, the top-level classification node in the knowledge graph of backup automatic transfer multi-mode protection test is matched to obtain the target logic subgraph corresponding to the set value of the backup automatic transfer mode; Traverse the protection logic function nodes, node dependency edges, and condition edges in the target logic subgraph to obtain the candidate test node set, node dependency relationship set, and condition edge set. Based on the control word semantic set and the logical condition set, the state of the condition edge set is determined to obtain the set of enabled test branches; Based on the set of enabled test branches, a target test node is selected from the set of candidate test nodes, and the target test node is sorted according to the set of node dependencies to obtain the target test item path; Based on the set of fixed values ​​and the set of control words semantics, the test parameter positions, state condition positions and expected criterion positions in the target test item path are instantiated to obtain an instantiated test sequence. The instantiated test sequence is compiled into a template to obtain the target test template.

[0010] Optionally, the step of determining the state of the conditional edge set based on the control word semantic set and the logical condition set to obtain the set of enabled test branches includes: Extract the protection function names and corresponding function activation / deactivation status from the control word semantic set to obtain a set of function activation / deactivation semantic tags; The set of semantic tags for enabling / disabling the protection function is associated with the set of conditional edges based on the name of the protection function, thus obtaining a set of conditional edge association relationships; Extract the branch activation conditions corresponding to each condition edge from the set of condition edges to obtain the set of branch activation conditions; Based on the function activation / deactivation status and the logical condition set, the branch activation condition set is parsed to obtain the branch condition status set; Based on the set of conditional edge associations, candidate test branches corresponding to each conditional edge are determined, resulting in a set of candidate test branches. If the branch condition status corresponding to the candidate test branch meets the corresponding branch activation condition, the candidate test branch is determined as the activated test branch, and the activated test branches are aggregated to obtain the activated test branch set.

[0011] Optionally, sending the target test template to the target test terminal, so that the target test terminal performs a simulated test on the target standby automatic transfer device based on the target test template, includes: The target test template is serialized to obtain the test template data packet; The test template data packet is subjected to integrity verification information generation and encryption processing to obtain an encrypted test task, which is then sent to the target test terminal through a secure communication channel. Send a time synchronization command to the target test terminal so that the target test terminal establishes a synchronized test time reference; The test initial state data collected by the target test terminal is received, and the test initial state data is compared with the initial state conditions in the target test template to obtain the initial state verification result. The test initial state data is the bus voltage data collected by the target test terminal through the acquisition module. If the initial state verification result indicates that the initial test state meets the initial state conditions, a test execution command is sent to the target test terminal. The test execution command causes the target test terminal to perform simulated test actions on the target standby automatic transfer device according to the synchronous test time base and the target test template. The simulated test actions include all instantiated test sequences in the target test template.

[0012] Furthermore, to achieve the above objectives, the present invention also proposes an automatic testing system for automatic switching devices (AS / RS) that applies the automatic testing method for AS / RS as described above, the automatic testing system for AS / RS comprising: The data processing module is used to preprocess the original setting file of the target backup automatic switching device and perform semantic parsing on the preprocessed original setting file to obtain a structured setting semantic object. The template construction module is used to perform operation mode matching and path selection on the structured fixed-value semantic object based on the pre-built backup self-starting multi-mode protection test knowledge graph to obtain the target test item path, and construct the target test template based on the target test item path. The backup self-starting multi-mode protection test knowledge graph takes the backup self-starting mode as the top-level classification node, and each top-level classification node corresponds to a logical subgraph of different backup self-starting modes. The testing module is used to send the target test template to the target test terminal, so that the target test terminal performs a simulated test action on the target standby automatic transfer device based on the target test template; The test verification module is used to receive the actual execution data returned by the target test terminal, and to perform logical verification and processing on the actual execution data to obtain the target verification report.

[0013] Furthermore, to achieve the above objectives, this application also proposes an automatic testing device for backup automatic transfer devices, the device comprising: a memory, a processor, and an automatic testing program for backup automatic transfer devices stored in the memory, the processor being used to run the automatic testing program for backup automatic transfer devices, the computer program being configured to implement the steps of the automatic testing method for backup automatic transfer devices as described above.

[0014] In addition, to achieve the above objectives, this application also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the automatic testing method for backup automatic switching devices as described above.

[0015] This invention preprocesses and semantically parses original value files, transforming inconsistent, fragmented, and manually interpretable value information into structured value semantic objects with clear field relationships and logical meanings. Based on a knowledge graph of multi-mode protection testing with backup automatic switching as the top-level classification node, it matches the operating mode and selects test paths for the structured value semantic objects. This allows protection functions, control conditions, action sequences, and mandatory inspection items under different backup automatic switching modes to be associated according to corresponding logical subgraphs, avoiding test logic gaps caused by relying solely on static parameter filling. A target test template is constructed based on the target test item path, transforming abstract protection logic into analog output, state switching, action monitoring, and judgment tasks that the target test terminal can directly execute, ensuring consistency between test parameters, execution sequence, and expected results. By receiving actual test execution data and performing logical verification and processing, it enables… The actual action status, action sequence, and contact feedback can form a closed-loop verification with the expected criteria in the target test template, and generate a target inspection report with a unified data structure accordingly. Because this invention utilizes structured fixed-value semantic objects to connect information parsing, knowledge reasoning, task instantiation, test execution, and result verification, the output of the previous stage directly constitutes the decision basis for the next stage, forming a closed-loop mechanism in which the test logic is dynamically driven by the fixed-value content. Thus, even when the backup automatic transfer device has multiple operating modes and complex control combinations, it can still automatically generate a complete test path that matches the actual configuration, and simultaneously reduce the logical mismatch, project omissions, and result fragmentation caused by manual translation. The complex multi-mode test logic generation process is completely transparent to the user, ensuring that even novice personnel can automatically generate a complete test plan covering all configuration operating modes, greatly improving the reliability, standardization level, and efficiency of on-site inspection. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of an automatic test device for a standby self-starting device in the hardware operating environment of the embodiment of the present invention; Figure 2 This is a flowchart illustrating the first embodiment of the automatic testing method for standby self-starting devices according to the present invention; Figure 3 A schematic diagram illustrating the interaction between a computer terminal and a target test terminal for executing the automatic testing method for standby self-starting devices according to the present invention; Figure 4 This is a flowchart illustrating the second embodiment of the automatic testing method for standby self-starting devices of the present invention; Figure 5 This is a structural block diagram of the first embodiment of the automatic testing system for standby self-starting devices of the present invention; Figure 6 This is a flowchart illustrating stage one of an embodiment of the automatic testing system for standby self-starting devices according to the present invention; Figure 7 This is a flowchart illustrating stage two of an embodiment of the automatic testing system for standby self-starting devices according to the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0020] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of an automatic test equipment for a standby self-starting device, which is part of the hardware operating environment involved in the embodiments of the present invention.

[0021] like Figure 1As shown, the automatic test equipment for the standby automatic switch device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and an input unit such as a keyboard; the user interface 1003 may also include standard wired and wireless interfaces. The network interface 1004 may optionally include standard wired and wireless interfaces (such as Wireless-Fidelity (Wi-Fi) interfaces). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0022] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the automatic test equipment for standby self-starting devices, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0023] like Figure 1 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and an automatic test program for the automatic switchgear.

[0024] exist Figure 1 In the automatic testing equipment for backup automatic transfer devices shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the automatic testing equipment for backup automatic transfer devices of the present invention can be set in the automatic testing equipment for backup automatic transfer devices. The automatic testing equipment for backup automatic transfer devices calls the automatic testing program for backup automatic transfer devices stored in the memory 1005 through the processor 1001 and executes the automatic testing method for backup automatic transfer devices provided in the embodiments of the present invention.

[0025] This invention provides an automatic testing method for standby automatic switching devices, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the automatic testing method for standby self-starting devices according to the present invention.

[0026] In this embodiment, the automatic testing method for the standby automatic switching device includes the following steps: Step S10: Preprocess the original setting file of the target standby automatic switching device, and perform semantic parsing on the preprocessed original setting file to obtain a structured setting semantic object.

[0027] It should be understood that the automatic testing method for automatic transfer switch (ATS) devices can be executed by an ATS automatic testing system. This system can be deployed on computer terminals, portable testing terminals, substation maintenance workstations, or intranet servers, and communicate with the target testing terminal. (Refer to...) Figure 3 , Figure 3 This diagram illustrates the interaction between the computer terminal deployed in the automatic self-starting test system of the present invention and the target test terminal. The following description uses the automatic self-starting test system as an example to illustrate this embodiment and the subsequent embodiments.

[0028] It should be noted that the target automatic transfer switch (ATS) device refers to an automatic backup power supply device that currently requires functional testing, periodic testing, or commissioning testing. The original setting sheet file refers to a data file recording the device model, automatic transfer method, control word, action setting value, action time, and functional configuration of the target ATS device. It can be an electronic document downloaded from the dispatch management system, a scanned document, or an image of the setting sheet obtained through image acquisition equipment.

[0029] A structured setpoint semantic object refers to a set of setpoint information organized according to a preset data structure. It may include device model, automatic switching mode, control word semantics, setpoint name, setpoint value, setpoint unit, and logical relationship between setpoints.

[0030] In its implementation, the automatic switching test system can receive electronic setting sheets exported from the dispatch management system, as well as images of paper setting sheets taken on-site. For electronic documents, text, table, and image layers can be extracted from each page; for scanned documents or captured images, tilt correction, noise filtering, brightness adjustment, and region segmentation can be performed. After preprocessing, information such as device model, automatic switching method, control words, undervoltage setting, overcurrent setting, and action delay can be extracted using text recognition models, table recognition models, semantic entity recognition models, or rule parsing programs. Furthermore, a correspondence between setting items and setting values ​​can be established based on field names, spatial locations, unit types, and contextual relationships.

[0031] For example, the automatic transfer switch (ATS) test system can associate "current setting of the overcurrent acceleration section of incoming line 2", "4.2", and "Ampere" into a single setting record. It can also interpret the device function and its on / off status corresponding to each control bit in the control word based on the model of the target ATS device. After parsing, the information can be encapsulated according to key-value structures, object structures, tree structures, or graph structures to form a structured setting semantic object.

[0032] Step S20: Based on the pre-built knowledge graph of backup self-starting multi-mode protection test, perform operation mode matching and path selection on the structured fixed value semantic object to obtain the target test item path, and construct the target test template based on the target test item path.

[0033] It should be noted that the knowledge graph for backup automatic transfer protection testing refers to graph-structured data used to represent the relationships between backup automatic transfer methods, device functions, test items, triggering conditions, execution order, test parameters, and expected results. The knowledge graph for backup automatic transfer protection testing uses backup automatic transfer methods as the top-level classification node, and each top-level classification node corresponds to a logical subgraph of different backup automatic transfer methods.

[0034] Among them, the top-level classification node refers to the upper-level node used to distinguish different backup self-transfer methods. The logic subgraph refers to the local graph structure corresponding to a backup self-transfer method and used to describe the test logic under that method. It can include functional nodes, condition nodes, parameter nodes, test nodes, and the associated edges between nodes.

[0035] It should be noted that the target test item path refers to the sequence of test items selected from the corresponding logic sub-graph and connected in logical order according to the actual operation mode and functional configuration of the target automatic transfer switch. The target test template refers to the task data that the target test terminal reads and executes, which may include analog voltage, analog current, switch status, signal duration, test step sequence, and expected action results.

[0036] It should be understood that by matching the corresponding logic subgraph with the backup automatic transfer mode as the entry point, the participation of irrelevant test logic in path selection can be reduced; by combining structured fixed-value semantic objects to select test nodes and conditional branches, the generated test content can be kept consistent with the actual configuration of the target backup automatic transfer device; by converting the test path into the target test template, test parameters, action sequence and result criteria can be unified, reducing omissions and sequence errors caused by manually arranging test items.

[0037] In some embodiments, the automatic transfer switch (ATS) test system can pre-construct a knowledge graph for multi-mode ATS protection testing. This graph uses "backup transfer mode" as the top-level classification node, with five independent subgraphs attached, each corresponding to a complete test logic path for one of the five standard operating modes: incoming line mutual transfer, segmented automatic transfer, and remote automatic transfer. Within each subgraph, nodes are "protection logic functions" (such as undervoltage judgment, backup transfer delay, closing logic, and accelerated section protection), and conditional edges are "parameter settings" and "control word semantics," forming a reasonable test logic network.

[0038] Step S30: Send the target test template to the target test terminal so that the target test terminal performs a simulated test on the target standby automatic switch device based on the target test template.

[0039] It should be noted that the target test terminal refers to the test equipment that can output analog electrical quantities or status signals to the target automatic transfer device and can collect the action results of the target automatic transfer device. It can be a relay protection tester, a portable intelligent tester, or an integrated secondary equipment test terminal.

[0040] Simulated test actions refer to the test process of establishing normal, fault, or abnormal states according to the target test template and monitoring the response of the target automatic transfer switch. It may include outputting simulated voltage, outputting simulated current, switching switch position signals, simulating communication status, and acquiring action contacts.

[0041] It should be understood that this embodiment, by having the target test terminal execute simulated test actions according to a unified template, can maintain the consistency of test input, execution order, and data collection time, and improve the test standardization between different testers and different test batches.

[0042] In its implementation, the automatic transfer switch (ATS) test system can send the target test template to the target test terminal via a wired communication interface, an internal communication network, or a secure wireless communication link. Upon receiving the target test template, the target test terminal can parse the test items, output parameters, duration, state switching conditions, and expected data collection content.

[0043] Before starting the test, the target test terminal can check the communication connection status, output channel status, and the initial status of the target backup automatic transfer device; after the check is completed, the test tasks in the target test template can be executed in sequence.

[0044] For example, in the incoming line switching test, the target test terminal can first output the analog voltage representing the normal operation of the two incoming lines, then reduce the analog voltage corresponding to the working incoming line, and simultaneously maintain the voltage of the backup incoming line at normal to simulate the working incoming line undervoltage state; subsequently, the target test terminal can collect the backup switch closing signal and action time.

[0045] For overcurrent acceleration function testing, a simulated fault current corresponding to the setpoint configuration can be applied after the backup switch is closed, and the protection action output signal can be collected. For remote backup automatic transfer testing, the position of the switch on the opposite side or the status of the communication link can also be simulated through the communication interface.

[0046] Step S40: Receive the actual execution data returned by the target test terminal, and perform logical verification and processing on the actual execution data to obtain the target inspection report.

[0047] It should be noted that the measured execution data refers to the data collected or generated by the target test terminal during the execution of simulated test actions, which may include analog output records, waveform data, switch quantity change records, action messages, action contact status, and action time. The target verification report is a structured report that records the test object, test configuration, test process, test data, and verification conclusions.

[0048] It should be understood that this embodiment can reduce the subjective differences caused by manual judgment of action results by associating the actual execution data with the expected criteria in the target test template; by verifying the action status, action sequence and action time at the same time, the working status of the target backup automatic switching device can be judged from two dimensions: functional logic and timing coordination.

[0049] In some embodiments, the target test terminal can send the test item identifier, output waveform, action contact change, action time and device message to the backup automatic transfer test system after completing each test item.

[0050] The backup self-starting test system can associate the actual test execution data with the corresponding test items in the target test template according to the test item identifier, and perform status verification, sequence verification and time verification respectively.

[0051] For example, the standby switch closing test after incoming line undervoltage can determine whether the undervoltage condition of the working incoming line is met, whether the standby switch operates under the set logic, and whether the actual operating time is within the allowable range. For the voltage transformer disconnection interlocking test, it can determine whether the standby switch closing action is interlocked after an abnormal voltage condition occurs. The automatic transfer switch (ATS) test system can generate qualified, unqualified, or pending verification conclusions based on various verification results, and write the test parameters, measured data, abnormal items, and inspection conclusions into the target inspection report.

[0052] In its implementation, the backup self-transfer test system receives the actual execution data returned by the target test terminal and associates the actual execution data with test item identifiers in the target test template to obtain a test item result set. The actual execution data includes raw waveform data, action messages, contact status data, and action timing data. It extracts the expected criteria set corresponding to each test item from the target test template. The expected criteria set includes expected action status, expected contact status, and expected action time range. It performs logical comparison and timing comparison between the test item result set and the expected criteria set to obtain logical verification results and timing verification results. Based on the logical verification results and the timing verification results, it determines the verification conclusion corresponding to each test item, obtaining a test item verification conclusion set. Finally, it classifies and encapsulates the test item verification conclusion set and the actual execution data according to the backup self-transfer method to obtain a target verification report.

[0053] This embodiment transforms unstructured setpoint information into structured setpoint semantic objects that can be used for automatic reasoning through preprocessing and semantic parsing, reducing human error in the process of setpoint reading and control logic interpretation. By using a multi-mode protection test knowledge graph for backup automatic switching, the operating mode, functional configuration, and setpoint parameters are used together for test path selection, enabling basic test items and conditional test items to be combined according to the actual configuration of the device. The test logic and target test terminal are connected through the target test template, ensuring consistency in test parameters, execution order, and data acquisition process. By performing logical and temporal verification on the actual execution data, the test conclusions are made to correspond with the original setpoint configuration and specific test items, thereby reducing parameter mismatch, test item omission, and data fragmentation in multi-operating mode testing, and improving the automation level, test completeness, result consistency, and data traceability of backup automatic switching device testing.

[0054] refer to Figure 4 , Figure 4 This is a flowchart illustrating the second embodiment of the automatic testing method for standby self-starting devices according to the present invention.

[0055] Based on the first embodiment described above, in this embodiment, step S10 further includes: Step S101: Perform image enhancement, distortion correction, and binarization processing on the original setpoint file of the target standby automatic switching device to obtain a standardized image.

[0056] In its implementation, the backup self-transfer test system extracts page images from electronic documents or receives fixed-value sheet images captured by scanning and camera equipment; performs grayscale conversion on the page images and enhances text edges using histogram equalization, local contrast adjustment, or sharpening; estimates the tilt angle based on the page outline, table line direction, or text baseline, and completes distortion correction through perspective transformation or rotation transformation; and then performs binarization based on global threshold, adaptive threshold, or local grayscale distribution.

[0057] For example, for tilted fixed-value images formed by on-site shooting, the four corners of the page can be detected first, then the rectangular page can be restored using four-point perspective transformation, and finally the page can be scaled to a uniform resolution to form a standardized image.

[0058] Step S102: Input the standardized image into the layout segmentation network and output the table region mask and the text description region mask.

[0059] It should be noted that page segmentation network refers to an image processing model that can identify different page types in a fixed-value single page according to pixels or regions.

[0060] A table area mask is a binary matrix or label matrix used to mark the table body, table boundaries, and internal areas of the table; a text description area mask is a binary matrix or label matrix used to mark titles, annotations, explanatory text, and other non-table text areas.

[0061] It should be understood that by separating the table area and the text description area, this embodiment can adopt appropriate text detection strategies for different layout structures, avoiding interference between table lines, cell boundaries and continuous explanatory text.

[0062] In its implementation, the backup self-deployment test system can pre-collect value sheet samples of different device models, different value sheet templates, and different image qualities, and perform pixel-level or region-level annotation on the table areas and text description areas in the samples; use the annotated samples to train the original semantic segmentation network, so that the network learns layout features such as table line distribution, cell arrangement, text density, and blank areas on the page.

[0063] After training, the standardized image is input into the layout segmentation network. The network outputs the classification probability of each pixel belonging to a table region, text description region, or background region. Then, table region masks and text description region masks are generated based on the classification probabilities. The layout segmentation network can adopt an encoder-decoder structure, a dilated convolution structure, or a multi-scale feature fusion structure. For example, it can be built based on the improved DeepLabV3Plus-CascadeASPP network.

[0064] Step S103: Perform text detection and character recognition based on the table region mask and the text description region mask to obtain a fixed-value single text sequence.

[0065] It should be noted that the setting sheet text sequence refers to a collection of text content organized according to the page position relationship. The text content may include device model, setting item name, setting value, unit, control word and explanatory text.

[0066] It should be understood that this embodiment uses a region mask to limit the text detection range, which can reduce false detections caused by table lines and background areas; and combines text position to complete character recognition and sequence recovery, which can preserve the spatial relationship between text content and the original layout.

[0067] In its implementation, the backup self-deployment test system locates the table area based on the table area mask and the non-table text area based on the text description area mask; it detects text boxes or text lines in each area and extracts the corresponding text sub-images from the standardized image; it performs character recognition on the text sub-images to obtain the text strings associated with each text position; and then sorts the text strings according to the page number, vertical coordinate, and horizontal coordinate.

[0068] For table areas, text can be organized according to the row and column order of cells; for text description areas, text can be organized according to the reading order from top to bottom and from left to right, ultimately forming a fixed-value text sequence.

[0069] Step S104: Input the fixed-value single text sequence into the entity recognition model and extract the entity triples in the fixed-value single text sequence.

[0070] It should be noted that an entity triple refers to associated data consisting of a setpoint, a setpoint value, and a setpoint unit <setpoint, value, unit>. It can also include device model entities, automatic transfer switch entities, and control word entities. For example, the overcurrent acceleration section current setpoint, 4.2, and amperes can form an entity triple.

[0071] In its implementation, the automatic transfer switch (ATS) test system utilizes a BERT-NER model pre-trained based on knowledge data from the power sector to identify entities in the text sequence of setting values. It automatically associates "overcurrent acceleration section current setting value," "4.2," and "A" into a triplet <setting value item, value, unit>. Specifically, the system prioritizes identifying the "standby switching mode" setting value, which is the core decision parameter for subsequent test logic selection.

[0072] Step S105: Decode the control words in the entity triples based on the preset protection model library to obtain the control word semantic set.

[0073] It should be noted that the preset protection model library refers to a data set that pre-stores different automatic transfer switch (ATS) device models, software versions, control word formats, control bit definitions, and functional meanings. Control word decoding refers to converting the control word into device function and function activation / deactivation status based on the control bit definition corresponding to the device model. The control word semantic set refers to a set of semantic tags consisting of multiple function names and their corresponding function activation / deactivation statuses.

[0074] In its implementation, the automatic switching test system identifies the device model and control word from the entity triplet, and queries the protection model library using the device model, software version, or setting sheet version as an index. Based on the control word length, encoding format, and control bit definition obtained from the query, the system converts the hexadecimal control word, binary control word, or character control word into a unified control bit sequence. Then, the values ​​of each control bit are mapped to the corresponding device function's active or deactivated state.

[0075] Furthermore, in order to accurately extract semantic information, step S105 above may include: Step S1051: Extract the target control word and device model from the entity triple.

[0076] It should be noted that the target control word refers to the coded data in the current target automatic transfer switch setting sheet used to indicate the functional activation / deactivation status of multiple devices. The device model refers to the data identifier that can distinguish the product type or series of the automatic transfer switch.

[0077] Step S1052: Extract the corresponding target control word definition table from the preset protection model library according to the device model.

[0078] It should be noted that the target control word definition table is a configuration data table that matches the device model and records the control word length, encoding method, control bit sequence number, device function, and status meaning. Selecting the control word definition table according to the device model can accommodate the differences in control word definitions between different manufacturers, different device series, and different software versions.

[0079] Step S1053: Determine the protection function corresponding to each bit of the target control word according to the target control word definition table.

[0080] It should be noted that a word position refers to a control position within a target control word that has an independent state meaning. Protection functions refer to the functions of devices corresponding to and participating in automatic switching action judgment, protection exits, interlocking, alarm, or control processing.

[0081] In its implementation, the automatic switching test system first determines the encoding base and word order of the target control word based on the target control word definition table; for hexadecimal control words, it can be converted into a binary control bit sequence; and then the corresponding protection function is found based on the word position number.

[0082] For example, one digit can correspond to the incoming line transfer function, another digit can correspond to the bus voltage transformer disconnection alarm function, and yet another digit can correspond to the overcurrent acceleration protection function.

[0083] Step S1054: Determine the function enable / disable status based on the target value of the target control word at each word position and the protection function corresponding to each word position.

[0084] It should be noted that the target value refers to the encoded value of the target control word at the corresponding word position. Function enable / disable status refers to whether the protection function is enabled or disabled, including both enabled and disabled states.

[0085] In its implementation, the automatic switchover test system reads the state mapping rules from the target control word definition table and matches the target value of each bit with the state mapping rule. If the target value corresponds to the activation code, the corresponding protection function is marked as activated; if the target value corresponds to the deactivation code, it is marked as deactivation. For functions that use multi-bit codes to represent the state, the corresponding bits can be combined first, and then the function state can be determined based on the combined value.

[0086] Step S1055: Combine the function enable / disable states corresponding to all word positions to obtain the control word semantic set.

[0087] It should be noted that the control word semantic set consists of multiple control word semantic tags. The control word semantic tags consist of protection functions and function activation / deactivation status, such as <incoming line mutual transfer function, activated>.

[0088] In one embodiment, after the standby automatic switching test system identifies the "control word" and its value "00011011", the system automatically triggers the control word decoding process: according to the device model The system retrieves the corresponding control word definition table from the system protection model library, expands "00011011" bit by bit, and generates a human-machine readable semantic set. For example: {<Incoming line mutual transfer function, enabled>, <Bus PT disconnection alarm function, enabled>, <Overcurrent acceleration section protection function, enabled>, ...}.

[0089] Step S106: Perform logical conflict verification on the entity triplet and the control word semantic set based on the preset relay protection coordination rules, and generate a structured fixed value semantic object by combining the entity triplet and the control word semantic set if the verification passes.

[0090] In its implementation, the system compares data based on built-in relay protection coordination rules. For example, rule R1 states: "If the backup transfer mode is 'incoming line mutual transfer,' then the 'incoming line mutual transfer enable' control word must be 'enabled.'" If a conflict is detected, the system generates a clear, highlighted alarm, prompting manual review.

[0091] Finally, the output is an absolutely accurate structured fixed-value semantic object. It contains the core decision parameter "backup mode", the complete control word semantic set, and all associated fixed-value items and their logical conditions.

[0092] This embodiment unifies the image quality of the fixed value unit through image enhancement, distortion correction, and binarization processing; preserves the structural differences between table and text areas through layout segmentation; establishes semantic associations between fixed value names, values, and units through text recognition and entity recognition; converts encoded information into function engagement / disengagement semantics through control word decoding; and filters out inconsistent configurations through logical conflict verification, thereby reducing the impact of image quality, layout differences, and control word interpretation deviations on the generation of subsequent test paths.

[0093] Based on the second embodiment described above, in the third embodiment of the automatic testing method for standby automatic switching devices of the present invention, step S103 further includes: Step S1031: Extract the table region image and the text description region image from the normalized image based on the table region mask and the text description region mask.

[0094] It should be noted that the table region image refers to the image content in the normalized image corresponding to the effective pixel positions of the table region mask; the text description region image refers to the image content in the normalized image corresponding to the effective pixel positions of the text description region mask.

[0095] In its implementation, the automatic test system performs connectivity analysis on the table area mask and the text description area mask to determine the bounding rectangle of each effective area. Based on the bounding rectangle, the corresponding image is cropped from the standardized image, and a preset pixel range is added outside the area boundary to prevent edge characters from being truncated. For adjacent or overlapping areas, they can be merged or separated according to the mask type and overlap ratio, ultimately forming the table area image and the text description area image.

[0096] Step S1032: Perform text detection on the table region image and the text description region image using a text detection model to obtain a set of text line coordinates.

[0097] It should be noted that the text line coordinate set refers to the position data of each text line in the standardized image coordinate system, including the text line coordinates of each cell in the table area and the text line coordinates in the text description area.

[0098] In its implementation, the backup self-casting test system collects fixed-value single-region images with text box annotations as training samples. It uses these samples to train a text detection model based on text proposal, pixel classification, or contour regression (for example, the text detection model could be a CTPN network). During training, the text box positions and text region labels are used as supervision information to enable the model to learn character edges, character spacing, and text line directions.

[0099] During runtime, the table region image and the text description region image are input into the text detection model respectively, and candidate text boxes are output. Then, candidate text boxes in the same text line are merged according to the overlap ratio, directional consistency and adjacency distance, and the local region coordinates are converted into standardized image coordinates to obtain the text line coordinate set.

[0100] Step S1033: Based on the set of text line coordinates, crop the standardized image to obtain a text line sub-image, and perform size normalization processing on the text line sub-image to obtain a text line image sequence.

[0101] In its implementation, the backup self-casting test system extracts text line sub-images from the standardized image based on the coordinates of each text line, appropriately expands the coordinate boundaries to preserve character strokes, scales the text line sub-images to a uniform height, and calculates the target width according to the original aspect ratio. For images with insufficient width, blank areas are filled; for images with a width exceeding the input limit, they are segmented according to the text direction. Subsequently, pixel normalization is performed, and each text line sub-image is sorted according to the text line coordinates to form a text line image sequence.

[0102] Step S1034: Input the text line image sequence into the pre-built character recognition model and output a set of text strings.

[0103] It should be noted that the character recognition model consists of a convolutional neural network feature extraction module, a bidirectional long short-term memory network module, and a connection-sequence classification module; The convolutional neural network feature extraction module is configured to perform local visual feature extraction on the text line image sequence to obtain a text line visual feature sequence. A bidirectional long short-term memory network module is configured to perform bidirectional context encoding on the visual feature sequence of the text line to obtain the context feature sequence of the text line. The connection time-series classification module is configured to perform character category prediction, duplicate character merging, and whitespace marker removal on the context feature sequence of the text line, and output a set of text strings.

[0104] It should be understood that convolutional neural networks can extract character strokes, edges, local structures, and character spacing features, and compress background pixels and repetitive image information, providing a compact visual representation for subsequent sequence modeling; bidirectional context encoding can use the relationship between characters to distinguish characters with similar shapes, and enhance the recognition stability of professional phrases, numerical units, and continuous control words; connection-based temporal classification can handle the problem of inconsistent image feature length and target character length, and avoid the cumulative impact of character segmentation errors on recognition results.

[0105] In its implementation, the backup self-transfer test system collects fixed-value single-line text images and corresponding text labels to construct training samples. The samples are processed with font changes, brightness changes, slight blurring, and scale changes to enhance the model's adaptability to different scanning qualities and printing formats. The training samples are input into the character recognition model, and the difference between the predicted sequence and the text label is calculated by connecting the temporal classification loss. Backpropagation is used to update the parameters of the convolutional neural network and the bidirectional long short-term memory network.

[0106] After the model is trained, the text line image sequence is input into the character recognition model in sequence to obtain a set of text strings corresponding to the coordinates of each text line.

[0107] The convolutional neural network feature extraction module extracts multi-layer visual features through convolution operations, non-linear activation, and downsampling; it compresses the image height dimension while maintaining the horizontal sequence dimension, and converts the feature map into a feature vector sequence column by column, thereby obtaining the text line visual feature sequence.

[0108] Then, the visual feature sequence of the text line is input into the forward long short-term memory network along the text direction, and into the backward long short-term memory network in reverse order. The bidirectional long short-term memory network module concatenates or merges the forward and backward hidden states at the same position to generate the text line context feature sequence. For example, when recognizing "current setpoint" and "time setpoint", adjacent characters and subsequent unit information can be used to reduce local character misjudgment.

[0109] Finally, the text line context feature sequence is input into the character classification layer (i.e., connected to the time series classification module) to obtain the character category probability corresponding to each time position. The character path with the higher probability is selected, and consecutive repeated characters are merged, whitespace markers are removed, and the text string is generated.

[0110] Step S1035: Sort and combine the text strings according to the set of text line coordinates to obtain a fixed-value single text sequence.

[0111] It should be understood that by restoring the text order by combining spatial location, the correspondence between the fixed value name and the fixed value value can be preserved, and the text in different cells or different paragraphs can be incorrectly spliced ​​together.

[0112] In its implementation, the backup self-deployment test system first groups the text strings according to the page number, and then determines the text line order according to the vertical coordinate. For text strings with similar vertical coordinates, they are sorted according to the horizontal coordinate. For table areas, the strings are mapped to the corresponding rows and columns based on the cell boundaries. For text description areas, the paragraph boundaries are determined based on the text line spacing and paragraph indentation. After sorting, the text content, coordinates, area type, and cell position are written together into the fixed value sheet text sequence.

[0113] Based on the above embodiments, in the fourth embodiment of the automatic testing method for standby automatic switching devices of the present invention, step S20 further includes: Step S201: Extract the standby self-initiated mode setting, control word semantic set, setting item set, and logical condition set from the structured setting semantic object.

[0114] It should be noted that the automatic transfer switch (ATS) mode setting refers to the setting value used to determine the operating logic type of the target ATS device. The setting value set refers to the data set consisting of each setting value name, setting value value, and setting value unit. The logic condition set refers to the conditional data formed based on setting value relationships, functional configurations, or verification rules.

[0115] Step S202: Match the top-level classification node in the knowledge graph of backup automatic transfer mode setting to obtain the target logic subgraph corresponding to the backup automatic transfer mode setting.

[0116] It should be noted that the target logic subgraph refers to a local structure of the knowledge graph that corresponds to the current backup self-deployment method and includes the test nodes and related relationships of this method.

[0117] In its implementation, the backup self-redirect test system matches the backup self-redirect mode setting with the mode name, mode code, or synonym name of the top-level classification node in the knowledge graph. After a successful match, it reads the target logical subgraph along the association edge between the top-level classification node and the logical subgraph.

[0118] For example, when the automatic transfer mode is sectional automatic transfer, the target logic sub-diagram may include test nodes such as bus undervoltage judgment, sectional switch closing sequence, and voltage transformer disconnection blocking.

[0119] Step S203: Traverse the protection logic function nodes, node dependency edges, and condition edges in the target logic subgraph to obtain the candidate test node set, node dependency relationship set, and condition edge set.

[0120] It should be noted that protection logic function nodes refer to graph nodes representing undervoltage judgment, backup delay, closing control, interlock judgment, alarm, or protection actions. Node dependency edges refer to associated edges representing the prerequisite relationships or execution order between test nodes. Conditional edges refer to branched associated edges that include function activation / deactivation, parameter status, or external conditions.

[0121] In its implementation, the backup self-starting test system performs a depth-first or breadth-first traversal starting from the starting node of the target logic subgraph, records the protection logic function nodes traversed as candidate test nodes, reads the predecessor, successor and parallel relationships between nodes to form a set of node dependencies, and reads the associated edges of the attached function activation / deactivation status, parameter thresholds or communication status to form a set of condition edges.

[0122] Step S204: Based on the control word semantic set and the logical condition set, perform state determination on the condition edge set to obtain the enabled test branch set.

[0123] It should be noted that enabling a test branch set refers to the set of branches that meet the conditions and are allowed to enter subsequent test nodes.

[0124] In its implementation, the backup self-starting test system matches the function name in the condition edge with the semantic label of the control word, and matches the parameter conditions in the condition edge with the set of logical conditions. When the function is in the activated state and the parameter conditions are met, the corresponding test branch is marked as enabled. When the function is in the deactivated state or the necessary parameters are missing, the corresponding test branch is not enabled.

[0125] Furthermore, in order to achieve dynamic and accurate changes in the condition test items according to the actual functional activation / deactivation status of the target standby automatic switching device, the above step S204 may include: Step S2041: Extract the protection function name and the corresponding function activation / deactivation status from the control word semantic set to obtain the function activation / deactivation semantic tag set.

[0126] It should be noted that the function activation / deactivation semantic tag set refers to a data set consisting of multiple protection function names and their corresponding activation or deactivation states. This embodiment converts the control word semantic set into a unified function activation / deactivation tag set, which allows the control word parsing results to adopt a consistent semantic expression with the knowledge graph condition edges.

[0127] Step S2042: Associate the set of function activation / deactivation semantic tags with the set of condition edges according to the name of the protection function to obtain the set of condition edge association relationships.

[0128] It should be noted that the set of conditional edge associations refers to the correspondence between the protection function name and the conditional edges in the knowledge graph. This embodiment, by establishing associations between semantic tags and conditional edges, can determine which test path branches are specifically controlled by the activation / deactivation status of each function.

[0129] Step S2043: Extract the branch activation conditions corresponding to each condition edge from the set of condition edges to obtain the set of branch activation conditions.

[0130] It should be noted that the branch activation condition refers to the logical constraint that allows entry into the test branch corresponding to the condition edge. It can be the implementation of a function, the existence of a fixed value, the validity of parameter relationships, the satisfaction of communication status, or a logical combination of multiple conditions.

[0131] In the specific implementation, the backup automatic transfer test system reads the condition fields and logical operators carried by each condition edge, and converts the condition fields into a unified condition expression; for example, the branch enable condition can be set to "overcurrent acceleration section protection function is enabled", or it can be set to "remote backup automatic transfer function is enabled and communication link simulation function is available"; each condition expression and the corresponding condition edge identifier are written into the branch enable condition set.

[0132] Step S2044: Based on the function activation / deactivation status and the logical condition set, perform condition state parsing on the branch activation condition set to obtain the branch condition state set.

[0133] It should be noted that the branch condition state refers to the satisfied or unsatisfied state calculated based on the current function enable / disable state and logical conditions.

[0134] Step S2045: Determine the candidate test branches corresponding to each condition edge based on the set of condition edge associations to obtain a set of candidate test branches.

[0135] It should be noted that candidate test branches refer to test path branches that are related to the semantics of the current control word and are waiting for condition judgment.

[0136] Step S2046: If the branch condition status corresponding to the candidate test branch meets the corresponding branch activation condition, the candidate test branch is determined as the activated test branch, and the activated test branches are aggregated to obtain the activated test branch set.

[0137] In the specific implementation, the backup self-starting test system reads the branch condition status corresponding to each candidate test branch one by one; if the branch condition status meets the corresponding branch activation condition, the candidate test branch is marked as an activated test branch; if the condition is not met, the candidate test branch is marked as a deactivated branch; then all activated test branches are summarized and deduplicated to form an activated test branch set.

[0138] Step S205: Select a target test node from the candidate test node set according to the set of enabled test branches, and sort the target test nodes according to the set of node dependencies to obtain the target test item path.

[0139] It should be noted that the target test node refers to the test node selected from the set of candidate test nodes that needs to be executed. The target test item path refers to the sequence of target test nodes arranged according to node dependencies.

[0140] In its implementation, the backup self-starting test system first selects the basic mandatory test nodes that are not controlled by the conditional edge, and then selects the conditional test nodes that enable the test branch association; it merges the two types of nodes into a target test node set; it performs topology sorting according to the node dependency relationship, and sorts the nodes at the same level according to the preset priority or the timing of the device action to obtain the target test item path.

[0141] Step S206: Instantiate the test parameter positions, state condition positions, and expected criterion positions in the target test item path according to the set of fixed values ​​and the set of control word semantics to obtain an instantiated test sequence.

[0142] It should be noted that the test parameter position refers to the parameter slot in the target test item path used to fill in voltage, current, frequency or duration; the state condition position refers to the condition slot used to fill in functional state, switch position or communication state; and the expected criterion position refers to the criterion slot used to fill in expected action state, contact state or action time range.

[0143] Step S207: Compile the instantiated test sequence into a template to obtain the target test template.

[0144] In its implementation, the backup automatic switching test system encodes the test item identifier, analog output, switch status, duration, acquisition channel, and expected criteria in the instantiated test sequence according to the data specifications supported by the target test terminal; after verifying the step sequence, parameter type, and required fields, it generates a structured target test template.

[0145] This embodiment selects the target logic subgraph by setting the running mode, obtains test nodes, dependencies, and conditional edges through graph traversal, filters valid test branches by controlling word semantics and logical conditions, forms test paths that satisfy the action sequence by sorting nodes, and generates executable test tasks through parameter instantiation and template compilation. This allows the test template to be adapted to self-starting mode, function activation / deactivation status, and specific set parameters, reducing project omissions and parameter mismatches caused by static templates. It effectively solves the problems of cumbersome and error-prone multi-mode logic arrangement and low efficiency of manual operation.

[0146] Based on the above embodiments, in the fifth embodiment of the automatic testing method for standby automatic switching devices of the present invention, step S30 further includes: Step S301: Serialize the target test template to obtain the test template data packet.

[0147] In its implementation, the backup automatic transfer test system encodes the task number, device identifier, test steps, output parameters, execution order, and expected criteria in the target test template according to the field order defined by the communication protocol; it converts character data, numerical data, and time data into a unified byte format and adds data length, protocol version, and template version to form a test template data packet.

[0148] Step S302: Generate and encrypt integrity verification information for the test template data packet to obtain an encrypted test task, and send the encrypted test task to the target test terminal through a secure communication channel.

[0149] In its implementation, the backup self-transfer test system uses a digest algorithm to calculate the digest value of the test template data packet and uses the digest value as integrity verification information; it uses a communication key to perform symmetric or hybrid encryption on the test template data packet to generate an encryption test task; and then sends the encryption test task to the target test terminal through a wired network, an intranet network, or a secure IoT gateway.

[0150] After the target test terminal completes decryption, it recalculates the digest value and compares the calculation result with the integrity verification information.

[0151] Step S303: Send a time synchronization command to the target test terminal so that the target test terminal establishes a synchronous test time reference.

[0152] It should be understood that the unified time reference in this embodiment can reduce the time deviation between different acquisition channels and different test steps, and improve the accuracy of action timing verification.

[0153] In its implementation, the standby self-starting test system sends a time synchronization command to the target test terminal. This command may include the time source type, synchronization mode, and allowable time deviation. The target test terminal can use BeiDou time synchronization, network time synchronization, a precise time protocol, or a local high-precision clock to complete time calibration and return the synchronization status and current time deviation. When the time deviation meets a preset synchronization condition, the system allows entry into the initial state verification phase. The preset synchronization condition can be that the time deviation does not exceed the maximum timing error allowed by the test task.

[0154] Step S304: Receive the initial test state data collected by the target test terminal, and compare the initial test state data with the initial state conditions in the target test template to obtain the initial state verification result.

[0155] It should be noted that the initial test state data is the bus voltage data collected by the target test terminal through the acquisition module; the initial state conditions refer to the range of bus voltage states or operating states specified in the target test template that allow the start of the test.

[0156] It should be understood that verifying the bus voltage before the test can prevent the test from continuing when the device is in an abnormal initial state and ensure that different test tasks have consistent starting conditions.

[0157] Step S305: If the initial state verification result indicates that the initial test state meets the initial state conditions, send a test execution command to the target test terminal.

[0158] In its implementation, when the initial state verification result is passed, the standby automatic transfer test system generates a test execution instruction containing the target test template identifier, start time, and execution mode, and sends the test execution instruction to the target test terminal; when the verification fails, it pauses the sending of the test execution instruction and outputs the corresponding bus voltage abnormality information for rewiring or adjustment of the test state.

[0159] Step S306: The test execution instruction causes the target test terminal to perform simulated test actions on the target standby automatic transfer device according to the synchronous test time base and the target test template. The simulated test actions include all instantiated test sequences in the target test template.

[0160] It should be noted that the simulated test action refers to the process by which the target test terminal outputs simulated voltage, simulated current, switch status or communication status according to the instantiated test sequence, and collects the action response of the target automatic transfer device.

[0161] In its implementation, the automatic backup switch test system triggers the target test terminal to read the target test template through the test execution command; the target test terminal executes each instantiated test sequence in sequence, outputs the analog voltage or analog current with corresponding amplitude, phase and duration, and switches the switch position input or communication status; at the same time, it collects the action output and signal contacts of the target automatic backup switch device.

[0162] For example, in the incoming line mutual transfer test, the normal state of the two incoming lines can be established first, and then the working incoming line undervoltage can be simulated and the closing action of the backup incoming line can be monitored; in the remote backup automatic transfer test, the communication link status and the position signal of the switch on the other side can be further simulated.

[0163] This embodiment improves the security of test task issuance, the consistency of test start state, and the accuracy of action timing acquisition by serializing the template transmission format, performing integrity verification and encryption, establishing a unified timing benchmark through time synchronization, constraining test start conditions through initial state verification, and controlling the target test terminal to continuously execute the instantiated test sequence through execution instructions.

[0164] Based on the above embodiments, in the sixth embodiment of the automatic testing method for standby automatic switching devices of the present invention, step S40 further includes: Step S401: Receive the actual execution data returned by the target test terminal, and associate the actual execution data with the test item identifier in the target test template to obtain the test item result set.

[0165] It should be noted that the actual test execution data includes raw waveform data, action messages, contact status data, and action timing data. The test item identifier is a unique identifier used to distinguish different test items and instantiated test sequences. The test item result set refers to the set of actual test data aggregated according to the test items.

[0166] Step S402: Extract the expected criterion set corresponding to each test item from the target test template.

[0167] It should be noted that the expected criterion set refers to the set of judgment conditions formed by the expected action state, expected contact state, and expected action time range corresponding to each test item. The expected action state can be action, no action, or lockout; the expected contact state can be closed, open, or state maintained.

[0168] It should be understood that extracting the expected criteria from the same target test template that generates the test input can keep the result judgment consistent with the test conditions and reduce the deviation caused by manually configuring the criteria separately.

[0169] In its implementation, the backup automatic transfer test system queries the target test template based on the test item identifier, reads the action exit criteria, signal contact criteria, and action time range in the corresponding test node, combines the criteria according to the test item, and checks for missing criteria or abnormal formats.

[0170] Step S403: Perform logical comparison and temporal comparison between the test item result set and the expected criterion set to obtain logical verification results and temporal verification results.

[0171] It should be understood that logical comparison can determine whether the device performs the correct action, while timing comparison can determine whether the action occurs within the specified time. Combining the two types of verification can avoid ignoring action delays or sequence abnormalities based solely on whether an action has been performed.

[0172] In its implementation, the automatic backup switching test system extracts the actual action exit, contact changes, and action time from the test project result set; compares the actual action state and actual contact state with the corresponding expected state item by item to generate a verification result that is logically consistent or logically inconsistent; and calculates the actual action delay using the fault trigger time or state switching time as the time starting point, and compares it with the expected action time range to generate a verification result that is timing qualified, timing out of limit, or timing missing.

[0173] Step S404: Determine the verification conclusions corresponding to each test item based on the logical verification results and the timing verification results, and obtain a set of test item verification conclusions; In its implementation, the backup self-starting test system combines the logic verification results and timing verification results according to preset conclusion rules. When both logic verification and timing verification meet the expected criteria, the test item is marked as qualified. When any mandatory criterion is not met, the test item is marked as unqualified, and the reason for the logic or timing abnormality is recorded. When the actual test data is missing, communication is interrupted, or the criteria cannot be completed, the test item is marked as pending review.

[0174] Step S405: Classify and encapsulate the test item inspection conclusion set and the actual test execution data according to the backup self-deployment method to obtain the target inspection report.

[0175] In its implementation, the automatic transfer switch (ATS) test system reads the ATS mode settings from the structured setpoint semantic object and groups the test item verification conclusion set and measured execution data according to categories such as incoming line mutual transfer, segmented automatic transfer, remote automatic transfer, bridge automatic transfer, or main transformer automatic transfer. It writes the device identifier, setpoint version, test time, test parameters, original waveform index, abnormal items, and verification conclusions into the target verification report. The report can be generated as an electronic document, structured data file, or web page report and is stored in association with the equipment file of the target ATS device.

[0176] This embodiment establishes a correspondence between multi-source measured data and test tasks by identifying test items, maintains consistency between test input and result judgment by extracting expected criteria, judges the correctness of actions and response time by logical verification and timing verification respectively, and forms a target inspection report by conclusion rules and classification encapsulation, thereby reducing the difference in manual judgment, preserving the correlation between test data, abnormal causes and backup self-deployment methods, and improving the consistency, traceability and subsequent call efficiency of inspection results.

[0177] Reference Figure 5 , Figure 5 This is a structural block diagram of the first embodiment of the automatic testing system for standby self-starting devices according to the present invention.

[0178] likeFigure 5 As shown, the automatic testing system for standby automatic switching devices proposed in this embodiment of the invention includes: The data processing module 10 is used to preprocess the original setting value file of the target backup automatic switching device and perform semantic parsing on the preprocessed original setting value file to obtain a structured setting value semantic object. The template construction module 20 is used to perform operation mode matching and path selection on the structured fixed-value semantic object based on the pre-built backup self-starting multi-mode protection test knowledge graph to obtain the target test item path, and construct the target test template based on the target test item path. The backup self-starting multi-mode protection test knowledge graph takes the backup self-starting mode as the top-level classification node, and each top-level classification node corresponds to a logical subgraph of different backup self-starting modes. Test module 30 is used to send the target test template to the target test terminal, so that the target test terminal performs a simulated test action on the target standby automatic transfer device based on the target test template; The test verification module 40 is used to receive the actual execution data returned by the target test terminal, and to perform logical verification and processing on the actual execution data to obtain a target verification report.

[0179] This embodiment transforms unstructured setpoint information into structured setpoint semantic objects that can be used for automatic reasoning through preprocessing and semantic parsing, reducing human error in the process of setpoint reading and control logic interpretation. By using a multi-mode protection test knowledge graph for backup automatic switching, the operating mode, functional configuration, and setpoint parameters are used together for test path selection, enabling basic test items and conditional test items to be combined according to the actual configuration of the device. The test logic and target test terminal are connected through the target test template, ensuring consistency in test parameters, execution order, and data acquisition process. By performing logical and temporal verification on the actual execution data, the test conclusions are made to correspond with the original setpoint configuration and specific test items, thereby reducing parameter mismatch, test item omission, and data fragmentation in multi-operating mode testing, and improving the automation level, test completeness, result consistency, and data traceability of backup automatic switching device testing.

[0180] The automatic testing system for backup automatic transfer switches provided in this application employs the automatic testing method for backup automatic transfer switches in the above embodiments, and can solve the technical problem of automatic testing for backup automatic transfer switches. Compared with the prior art, the beneficial effects of the automatic testing system for backup automatic transfer switches provided in this application are the same as the beneficial effects of the automatic testing method for backup automatic transfer switches provided in the above embodiments, and other technical features of the automatic testing system for backup automatic transfer switches are the same as the features disclosed in the methods of the above embodiments, and will not be repeated here.

[0181] In one embodiment of the automatic testing system for standby automatic transfer switches (ATSPS) described in this invention, a 110kV substation is used as the object. The automatic testing system for ATSPS performs periodic inspections on an ATSPS model "NSR3641C". This device is configured with five operating modes: incoming line mutual transfer, sectional automatic transfer, and remote standby automatic transfer, each of which needs to be verified individually.

[0182] The input data for the automatic test system of the standby automatic transfer switch is a PDF electronic document or JPG image containing 6 pages of text and images, downloaded from the OMS system or obtained by on-site photography. The content includes the device model "NSR3641C", the core setting "standby transfer mode = incoming line mutual transfer", the control string "00011011", and "2# incoming line overcurrent acceleration section current setting = 4.2A, time = 0.3s", etc.

[0183] The specific processing flow of the automatic testing system for standby automatic switching devices includes two stages: Phase 1: Intelligent recognition and deep semantic understanding of fixed value sheets, refer to Figure 6 , Figure 6 This is a flowchart illustrating the first phase of an automatic testing system for standby automatic switching devices.

[0184] This phase addresses the issues of "human eye reading" and "human brain understanding" in existing technologies, providing an accurate data foundation for subsequent multi-method automatic matching.

[0185] Step A1: Image Preprocessing and Layout Analysis System Acquisition Then, image enhancement, distortion correction, and binarization are performed to obtain a standardized image. Next, a page segmentation model based on the improved DeepLabV3Plus-CascadeASPP network is invoked, which uses... For input, output two masks: a table area. and text description area This is to achieve image-text separation and precise positioning.

[0186] Step A2: Text detection, recognition, and initial extraction: exist The text in the image is then processed using a CTPN network to detect the coordinates of text lines within each cell. Each text line subimage is then fed into an end-to-end recognition network combining CNN, BiLSTM, and CTC to output the corresponding text string. In this way, the text in the image is transformed into an editable text sequence.

[0187] Step A3: Semantic parsing and error correction based on expert rules: This is a crucial step beyond simple OCR, aiming to transform text sequences into relay protection semantic objects that computers can "understand." The system is configured with an expert rule engine for automatic transfer switch (ATS) setting sheets, and the workflow is as follows: Step A3.1, Key-Value Pair Extraction and Association: Using a BERT-NER model fine-tuned for the power industry, entities in the text sequence are identified, and "overcurrent acceleration section current setting", "4.2", and "A" are automatically associated as triples <setting item, value, unit>. In particular, the system prioritizes identifying the "standby mode" setting, which is the core decision parameter for subsequent test logic selection.

[0188] Step A3.2, Deep Decoding of Control Word: After recognizing the "control word" and its value "00011011", the system automatically triggers the control word decoding process: based on the device model... The system retrieves the corresponding control word definition table from the system protection model library, expands "00011011" bit by bit, and generates a human-machine readable semantic set. For example: {<Incoming line mutual transfer function, enabled>, <Bus PT disconnection alarm function, enabled>, <Overcurrent acceleration section protection function, enabled>, ...}.

[0189] Step A3.3, Logical Conflict Verification: The system compares data according to the built-in relay protection coordination rules. For example, rule R1: "If the backup transfer mode is 'incoming line mutual transfer', then the 'incoming line mutual transfer enable' control word must be 'enabled'." If a conflict is found, the system generates a clear highlighted alarm, prompting manual review.

[0190] Ultimately, this stage outputs an absolutely accurate structured fixed-value semantic object. . Includes the core decision parameter "backup deployment method" and a complete set of control word semantics. And all associated fixed-value items and their logical conditions.

[0191] Phase Two: One-click generation and execution of logic-driven test tasks for multiple operating modes, refer to... Figure 7 , Figure 7 This is a flowchart illustrating the second phase of the automatic testing system for standby automatic switching devices.

[0192] This phase addresses the problems of "cumbersome and error-prone multi-mode logic arrangement" and "low efficiency of manual operation" in existing technologies, which is another core innovation of this solution.

[0193] Step A4: Automatic generation of test templates based on multi-method knowledge graphs: The system constructs a knowledge graph for multi-mode protection testing. This graph uses "backup transfer mode" as the top-level classification node, with five independent subgraphs attached, each corresponding to a complete test logic path for one of the five standard operating modes: incoming line mutual transfer, segmented automatic transfer, and remote backup automatic transfer. Within each subgraph, nodes are "protection logic functions" (such as undervoltage judgment, backup transfer delay, closing logic, and accelerated section protection), and conditional edges are "parameter settings" and "control word semantics," forming a reasonable test logic network.

[0194] Step A4.1, Automatic Matching and Path Selection: The system reads... The core value "Backup Transfer Mode = Incoming Line Interconnection" is automatically matched in the top-level node of the knowledge graph, precisely navigating to the "Incoming Line Interconnection Mode Test" subgraph. The system extracts all mandatory test item paths under this mode from this subgraph, ensuring comprehensive test coverage. If the device is configured with remote backup automatic transfer mode, the system automatically selects the remote backup automatic transfer subgraph, which additionally includes specialized test nodes such as "Communication Link Status Confirmation," "Simulation of Opposite Side Switch Position Signal," and "Cross-Site Timing Coordination."

[0195] Step A4.2, instantiate the test logic driven by the control word: Basic parameter mapping: For each test point in the path (such as "1# incoming line undervoltage judgment"), the system will... Enter the corresponding set value in the test template. For example, enter "1# incoming line undervoltage set value = 30V" in the expected action parameter field of the test template.

[0196] Conditional branch injection: system read The semantics of <Overcurrent Acceleration Section Protection Function, Activated> trigger the conditional edge in the knowledge graph. After the "Backup Automatic Transfer Closing" test sequence is completed, the test step of "Applying a fault current > 4.2A for 0.3 seconds, with the expected overcurrent acceleration section protection action and tripping" is automatically added. If the control word is "Exit", this step is automatically deleted.

[0197] Step A4.3, Template File Compilation: All the instantiated parameters and defined logical sequences described above are compiled into a platform-independent, executable XML format test template file. .

[0198] The standardized coverage description of the five operating modes in this embodiment is as follows: In this embodiment, the self-submitting multi-mode knowledge graph has pre-modeled the test logic for the following five standard operating modes, as shown in Table 1 below: Table 1. Test Logic Table for Automatic Start-up Switching Mode The system according to The value of "backup deployment method" will automatically select the corresponding subgraph to generate a test template, eliminating the need for manual arrangement for each method.

[0199] Step A5: Secure Distribution and "One-Click Execution": Step A5.1, Secure Distribution: The computing terminal, through a secure IoT gateway with a built-in national cryptographic chip, will... Encrypted transmission to portable smart tester.

[0200] Step A5.2, Synchronization and Readiness: The tester uses the BeiDou time synchronization module to complete time synchronization and uses its own acquisition module to read the bus voltage in real time, and... The initial normal state is compared to ensure that the test preconditions are met.

[0201] Step A5.3, One-click automatic test: The operator presses "Start Test". The tester begins automatic execution. All test sequences defined in the document include: normal state simulation under various modes, fault state triggering and action monitoring, timing coordination verification, communication link state simulation and cross-site signal interaction verification (remote backup automatic transfer mode).

[0202] Step A6: Automatic Result Judgment and Reporting Closure: Step A6.1, item-by-item judgment: The tester will compare the execution result of each test step with... The criteria within the system are compared logically and temporally to arrive at a "qualified / unqualified" conclusion.

[0203] Step A6.2, Data and Report Upload: All raw waveform data, action messages, and judgment conclusions are packaged and automatically uploaded back to the intranet cloud platform. The platform starts the report microservice, generates an unalterable standardized inspection report, and archives it according to the operating mode, permanently associating it with the automatic transfer switch as equipment status data.

[0204] The output includes: Main result: A standardized inspection report that is automatically generated, automatically uploaded, and categorized by operating mode.

[0205] Data Results: A structured inspection data archive covering all configuration and operating modes, which can be accessed by the condition-based maintenance system.

[0206] It should be noted that the beneficial effects of this invention include: 1. Deep semantic analysis and logical verification method for "control words" in standby automatic switching settings: Unlike existing technologies that simply store control words as hexadecimal strings, this invention features an innovative "adaptive decoding" and "associative logical conflict verification" mechanism for control words. It automatically associates the control word with a bit definition table based on the device model, decoding it into a series of semantic tags for <function, enable / disable>, and performing cross-field logical consistency verification with other core settings.

[0207] Beneficial effects: This method represents a qualitative leap from "recording data" to "understanding logic." It functions like a built-in expert review system, proactively identifying potential hidden errors in the setting sheet itself (such as contradictions between "backup method" and "control word") before test execution. This avoids omissions and logical errors caused by misunderstandings or oversights when manually arranging test sequences for each method, thus constructing the first and most fundamental line of defense for automated testing.

[0208] 2. Automatic planning and generation technology for self-initiated test tasks based on multi-method knowledge graphs and driven by control words: Existing technologies generate static test templates based on manually set parameter lists. This invention proposes a dynamic, multi-mode knowledge graph-based automatic planning method. It uses "backup deployment mode" as the top-level decision node, incorporating complete test logic subgraphs for five standard operating modes: incoming line mutual deployment, segmented self-deployment, and remote backup self-deployment. Control word semantics serve as the key "switch" driving the automatic orchestration of test branches. In particular, for the remote backup self-deployment mode, the system automatically adds specialized test nodes such as communication link status simulation and cross-site timing coordination, which are easily overlooked in manual orchestration.

[0209] Beneficial effects: It truly overcomes the automation challenges of "multiple methods, complex logic, and easy omissions" in automatic start-up testing. Regardless of the operating mode configured for the device, the system can automatically match the corresponding standardized test path from the knowledge graph, ensuring comprehensive and complete coverage of test items for all five modes. It makes the complex multi-mode test logic generation process completely transparent to the user, ensuring that even novice personnel can automatically generate complete test plans covering all configuration operating modes, greatly improving the reliability, standardization, and efficiency of on-site inspection.

[0210] 3. A one-click intelligent closed-loop testing system architecture that connects the entire process of "determination-testing-evaluation-storage": Unlike traditional testing instruments that exist in isolation as single-point tools, this invention achieves seamless integration and secure interconnection of "cloud-based value acquisition", "on-site intelligent analysis", "one-click execution of multiple terminal methods", and "cloud-based method-based classification reporting" from a system architecture perspective.

[0211] Beneficial effects: It completely breaks down data silos. The system architecture enables unified management of inspection data for the entire lifecycle and all operating modes of the automatic transfer switch (ATS) device. This ensures that each field test is not only a current task but also a structured data injection for assessing the future state of the equipment. This represents progress from "tool empowerment" to "system innovation," providing a solid data foundation for condition-based maintenance of protective equipment and pre-control of power grid risks.

[0212] Furthermore, embodiments of the present invention also propose a computer-readable storage medium storing an automatic test program for a backup automatic transfer device, wherein when the automatic test program for the backup automatic transfer device is executed by a processor, the automatic test program for the backup automatic transfer device implements the steps of the automatic test method for the backup automatic transfer device as described above.

[0213] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0214] The aforementioned computer-readable storage medium may be included in an automatic test apparatus for backup automatic switching devices; or it may exist independently and not be assembled into an automatic test apparatus for backup automatic switching devices.

[0215] Furthermore, this invention also proposes a computer program product, including an automatic test program for backup automatic transfer devices, wherein when the automatic test program for backup automatic transfer devices is executed by a processor, it implements the steps of the automatic test method for backup automatic transfer devices as described above.

[0216] The specific implementation of the computer program product of the present invention is basically the same as the embodiments of the automatic testing method for standby self-starting devices described above, and will not be repeated here.

[0217] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0218] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0219] In addition, for technical details not described in detail in this embodiment, please refer to the automatic testing method for standby self-starting devices provided in any embodiment of the present invention, which will not be repeated here.

[0220] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0221] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0222] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0223] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. An automatic testing method for standby automatic switching devices, characterized in that, The automatic testing method for standby automatic switching devices includes: The original setting file of the target standby automatic switching device is preprocessed, and the preprocessed original setting file is semantically parsed to obtain a structured setting semantic object. Extract the backup self-starting mode setting, control word semantic set, setting item set, and logical condition set from the structured setting semantic object; match the backup self-starting mode setting with the top-level classification node in the backup self-starting multi-mode protection test knowledge graph to obtain the target logical subgraph corresponding to the backup self-starting mode setting, wherein the backup self-starting mode is the top-level classification node, and each top-level classification node corresponds to a logical subgraph of different backup self-starting modes; traverse the protection logic function nodes, node dependency edges, and condition edges in the target logical subgraph to obtain the candidate test node set, node dependency relationship set, and condition set. Edge set; based on the control word semantic set and the logical condition set, the condition edge set is state-determined to obtain an enabled test branch set; target test nodes are selected from the candidate test node set according to the enabled test branch set, and the target test nodes are sorted according to the node dependency set to obtain a target test item path; the test parameter positions, state condition positions, and expected criterion positions in the target test item path are instantiated according to the fixed value item set and the control word semantic set to obtain an instantiated test sequence; the instantiated test sequence is template-compiled to obtain a target test template; The target test template is sent to the target test terminal, so that the target test terminal performs a simulated test on the target standby automatic transfer device based on the target test template. The system receives the actual execution data returned by the target test terminal, performs logical verification and processing on the actual execution data, and obtains the target inspection report.

2. The automatic testing method for standby automatic switching devices as described in claim 1, characterized in that, The process of preprocessing the original setting file of the target standby automatic switching device and performing semantic parsing on the preprocessed original setting file to obtain a structured setting semantic object includes: The original setpoint file of the target automatic switching device is subjected to image enhancement, distortion correction and binarization processing to obtain a standardized image; The standardized image is input into the layout segmentation network, which outputs a table region mask and a text description region mask. Text detection and character recognition are performed based on the table region mask and the text description region mask to obtain a fixed-value single text sequence; The fixed-value single text sequence is input into the entity recognition model to extract the entity triples in the fixed-value single text sequence; The control words in the entity triples are decoded based on a preset protection model library to obtain a control word semantic set; Logical conflict verification is performed on the entity triplet and the control word semantic set based on the preset relay protection coordination rules. If the verification passes, a structured fixed-value semantic object is generated by combining the entity triplet and the control word semantic set.

3. The automatic testing method for standby automatic switching devices as described in claim 2, characterized in that, The text detection and character recognition based on the table region mask and the text description region mask, to obtain a fixed-value single text sequence, includes: Based on the table region mask and the text description region mask, extract the table region image and the text description region image from the standardized image; Text detection is performed on the table area image and the text description area image using a text detection model to obtain a set of text line coordinates. The set of text line coordinates includes the text line coordinates of each cell in the table area and the text line coordinates in the text description area. The standardized image is cropped based on the set of text line coordinates to obtain text line sub-images, and the size of the text line sub-images is normalized to obtain a text line image sequence. The text line image sequence is input into a pre-built character recognition model, which outputs a set of text strings; Based on the set of text line coordinates, the set of text strings is sorted by position and the text is combined to obtain a fixed-value single text sequence; The character recognition model consists of a convolutional neural network feature extraction module, a bidirectional long short-term memory network module, and a connection-sequence classification module. The convolutional neural network feature extraction module is configured to perform local visual feature extraction on the text line image sequence to obtain a text line visual feature sequence. The bidirectional long short-term memory network module is configured to perform bidirectional context encoding on the visual feature sequence of the text line to obtain the context feature sequence of the text line. The connection time-series classification module is configured to perform character category prediction, duplicate character merging, and whitespace marker deletion on the text line context feature sequence, and output a set of text strings.

4. The automatic testing method for standby automatic switching devices as described in claim 2, characterized in that, The process of decoding the control words in the entity triples based on a preset protection model library to obtain a control word semantic set includes: Extract the target control word and device model from the entity triple; According to the device model, the corresponding target control word definition table is extracted from the preset protection model library; The protection function corresponding to each bit of the target control word is determined according to the target control word definition table; The function enable / disable status is determined based on the target value of the target control word at each word position and the protection function corresponding to each word position. The function activation / deactivation status corresponding to all characters is combined to obtain a control word semantic set. The control word semantic set consists of multiple control word semantic tags, and the control word semantic tags consist of protection functions and function activation / deactivation status.

5. The automatic testing method for standby automatic switching devices as described in claim 1, characterized in that, The step of determining the state of the conditional edge set based on the control word semantic set and the logical condition set to obtain the enabled test branch set includes: Extract the protection function names and corresponding function activation / deactivation status from the control word semantic set to obtain a set of function activation / deactivation semantic tags; The set of semantic tags for enabling / disabling the protection function is associated with the set of conditional edges based on the name of the protection function, thus obtaining a set of conditional edge association relationships; Extract the branch activation conditions corresponding to each condition edge from the set of condition edges to obtain the set of branch activation conditions; Based on the function activation / deactivation status and the logical condition set, the branch activation condition set is parsed to obtain the branch condition status set; Based on the set of conditional edge associations, candidate test branches corresponding to each conditional edge are determined, resulting in a set of candidate test branches. If the branch condition status corresponding to the candidate test branch meets the corresponding branch activation condition, the candidate test branch is determined as the activated test branch, and the activated test branches are aggregated to obtain the activated test branch set.

6. The automatic testing method for standby automatic switching devices as described in claim 1, characterized in that, Sending the target test template to the target test terminal, so that the target test terminal performs a simulated test on the target standby automatic transfer device based on the target test template, includes: The target test template is serialized to obtain the test template data packet; The test template data packet is subjected to integrity verification information generation and encryption processing to obtain an encrypted test task, which is then sent to the target test terminal through a secure communication channel. Send a time synchronization command to the target test terminal so that the target test terminal establishes a synchronized test time reference; The test initial state data collected by the target test terminal is received, and the test initial state data is compared with the initial state conditions in the target test template to obtain the initial state verification result. The test initial state data is the bus voltage data collected by the target test terminal through the acquisition module. If the initial state verification result indicates that the initial test state meets the initial state conditions, a test execution command is sent to the target test terminal. The test execution command causes the target test terminal to perform simulated test actions on the target standby automatic transfer device according to the synchronous test time base and the target test template. The simulated test actions include all instantiated test sequences in the target test template.

7. An automatic testing system for standby automatic switching devices, characterized in that, The system employs the automatic testing method for standby automatic switching devices as described in any one of claims 1 to 6, and the system comprises: The data processing module is used to preprocess the original setting file of the target backup automatic switching device and perform semantic parsing on the preprocessed original setting file to obtain a structured setting semantic object. The template construction module is used to perform operation mode matching and path selection on the structured fixed-value semantic object based on the pre-built backup self-starting multi-mode protection test knowledge graph to obtain the target test item path, and construct the target test template based on the target test item path. The backup self-starting multi-mode protection test knowledge graph takes the backup self-starting mode as the top-level classification node, and each top-level classification node corresponds to a logical subgraph of different backup self-starting modes. The testing module is used to send the target test template to the target test terminal, so that the target test terminal performs a simulated test action on the target standby automatic transfer device based on the target test template; The test verification module is used to receive the actual execution data returned by the target test terminal, and to perform logical verification and processing on the actual execution data to obtain the target verification report.

8. An automatic testing device for a standby self-starting device, characterized in that, The automatic testing equipment for backup automatic transfer devices includes: a memory, a processor, and an automatic testing program for backup automatic transfer devices stored in the memory. The processor is used to run the automatic testing program for backup automatic transfer devices, and the automatic testing program for backup automatic transfer devices is configured to implement the automatic testing method for backup automatic transfer devices as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an automatic test program for a backup automatic transfer device, which, when executed by a processor, implements the automatic test method for a backup automatic transfer device as described in any one of claims 1 to 6.

Images