Relay protection verification method and device, computer device and program product

Abstract

The application relates to a relay protection verification method and device, computer equipment and a program product. A specification text of a relay protection device is acquired; action logic information in the specification text is analyzed, condition information and corresponding action information in the action logic information are extracted, and a condition attribute set is obtained; a verification decision tree is constructed based on the condition attribute set; a corresponding target verification path is matched in the verification decision tree according to a verification requirement of the relay protection device; target condition information in the target verification path is taken as a verification condition, target action information in a leaf node corresponding to the target condition information is taken as an expected result, the operation of the relay protection device is verified, and a corresponding verification conclusion report is obtained. In this way, the artificial dependence is reduced, the completeness and efficiency of the verification are improved, and the relay protection devices of different manufacturers can be conveniently adapted.

Description

Technical Field

[0001] This application relates to the field of power safety technology, and in particular to a verification method, apparatus, computer equipment, and program product for relay protection. Background Technology

[0002] As the core barrier for the safe and stable operation of power systems, the precise execution of relay protection devices directly determines the isolation speed and system recovery capability during grid faults. During device development, grid connection testing, and daily operation and maintenance, verifying the operating logic is a crucial step in ensuring the reliability of protection functions.

[0003] Currently, protection logic is primarily embedded in manufacturer manuals in unstructured natural language. Testers must rely on their personal experience to manually parse this text, identify condition combinations, and design test cases. However, as the complexity of smart grids increases, the condition branches of protection logic grow exponentially. The manual parsing process is not only inefficient but also prone to missing key combinations in boundary scenarios with multiple overlapping conditions, reducing the completeness and reliability of verification.

[0004] Furthermore, there are significant differences in the implementation details of relay protection devices from different manufacturers, which makes it difficult to adapt to general testing methods. Verification schemes must be customized for each manufacturer, which greatly increases the complexity and time cost of testing. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, device, computer equipment, and program product for verifying relay protection to address the above-mentioned technical problems. This would reduce reliance on manual labor, improve the completeness and efficiency of verification, and facilitate compatibility with relay protection devices from different manufacturers.

[0006] Firstly, this application provides a method for verifying relay protection, including:

[0007] Obtain the instruction manual for the relay protection device;

[0008] The action logic information in the instruction manual text is parsed, and the condition information and corresponding action information in the action logic information are extracted to obtain the condition attribute set.

[0009] Based on the set of conditional attributes, a verification decision tree is constructed. The verification decision tree includes multiple verification paths extending from the root node. Each verification path includes multiple branch nodes and corresponding leaf nodes. Branch nodes map conditional information, and leaf nodes map action information.

[0010] Based on the verification requirements of the relay protection device, the corresponding target verification path is matched in the verification decision tree;

[0011] The target condition information in the target verification path is used as the verification condition, and the target action information in the leaf node corresponding to the target condition information is used as the expected result to verify the operation of the relay protection device and obtain the corresponding verification conclusion report.

[0012] In one embodiment, the step of parsing the action logic information in the specification text and extracting the condition information and corresponding action information from the action logic information to obtain the condition attribute set includes:

[0013] Natural language processing technology is used to identify paragraphs in the instruction manual that describe the action logic and extract key information from them. The key information is used to represent the action logic information of the relay protection device. Natural language processing technology includes named entity recognition or domain-based regular expression matching.

[0014] Extract condition information and corresponding action information from key information; among them, condition information includes action conditions and interlocking conditions. Action conditions include at least fault type, electrical quantity threshold, delay time and directional element status. Interlocking conditions include at least control word status, soft pressure plate status and external abnormal status. Action information includes at least tripping, alarm and reclosing.

[0015] The extracted conditional information is matched and associated with the corresponding action information to obtain a conditional action mapping dataset.

[0016] A conditional attribute library is constructed based on the conditional action mapping dataset to structure the conditional and action information in the dataset and obtain a set of conditional attributes.

[0017] In one embodiment, the step of constructing a standard conditional attribute library based on the conditional action mapping dataset to structure the conditional and action information in the dataset and obtain a set of conditional attributes includes:

[0018] Establish a condition attribute library; the condition attribute library shall include at least the following: a fault type attribute library, which is discrete and includes single-phase grounding, phase-to-phase short circuit, and three-phase short circuit; an electrical quantity threshold attribute library, which is continuous and includes overcurrent setting, impedance setting, and zero-sequence voltage setting; a delay time attribute library, which is numerical and includes action delay; a control word status attribute library, which is Boolean and includes "on" and "off"; and a soft pressure plate status attribute library, which is Boolean and includes "on" and "off".

[0019] The condition information in the condition action mapping dataset is categorized according to the categories of the condition attribute library, and the condition attribute library is labeled with data type, physical quantity type, unit, variable symbol and manufacturer mapping information to obtain the condition attribute set.

[0020] In one embodiment, the step of constructing a validation decision tree based on a set of conditional attributes includes:

[0021] A labeled multidimensional dataset is constructed by using the conditional attributes in the conditional attribute set as feature dimensions and the corresponding action information as classification labels.

[0022] Iterate through the conditional attributes in the multidimensional dataset and calculate the information gain of multiple conditional attributes;

[0023] The conditional attribute with the largest information gain is selected as the current split node, and the process continues until the action information of each conditional attribute in the multidimensional dataset is uniquely determined or the conditional attributes are exhausted, thus forming a validation decision tree.

[0024] In one embodiment, the step of traversing the conditional attributes in the dataset and calculating the information gain of each conditional attribute includes:

[0025] The information entropy of the multidimensional dataset is calculated based on the frequency of occurrence of various types of action information in the multidimensional dataset; whereby the information entropy is used to represent the uncertainty measure of the multidimensional dataset.

[0026] The multidimensional dataset is divided into multiple dimensional subsets based on the values ​​of the conditional attributes. The information entropy of the multiple dimensional subsets is calculated, and the sum is weighted according to the number of samples in each dimensional subset to obtain the conditional entropy of the conditional attribute.

[0027] The information gain of a conditional attribute is obtained by subtracting the conditional entropy of the conditional attribute from the information entropy of the dataset.

[0028] In one embodiment, the verification requirements include a full-path coverage strategy, a boundary condition priority strategy, or a specified path strategy. The step of matching the corresponding target verification path in the verification decision tree based on the verification requirements of the relay protection device includes:

[0029] When using the full path coverage strategy, all paths from the root node to each leaf node of the verification decision tree are traversed to generate the target verification path set.

[0030] Alternatively, when using a boundary condition priority strategy, the path containing the combination of blocking conditions and the electrical quantity threshold boundary value should be selected as the target verification path.

[0031] Alternatively, when a specified path strategy is used, the corresponding verification path can be located as the target verification path based on the combination of input condition attributes.

[0032] In one embodiment, the steps of using the target condition information in the target verification path as the verification condition and the target action information in the leaf node corresponding to the target condition information as the expected result to verify the operation of the relay protection device and obtain the corresponding verification conclusion report include:

[0033] If the condition attribute in the target verification path is Boolean, then set the state of the corresponding control word and soft pressure plate of the relay protection device;

[0034] If the condition attribute in the target verification path is continuous, then set the incremental parameters of the relay protection tester; the incremental parameters shall include at least the magnitude and phase angle of the fault phase voltage and fault phase current.

[0035] Set timing parameters; where the timing parameters include at least the pre-fault state holding time and the fault state duration;

[0036] The control relay protection tester performs physical input according to the input parameters, pre-state and timing parameters;

[0037] Monitor the real-time action signals, operating data signals, and switch on / off signals fed back by the relay protection device, determine and calculate the actual action behavior and actual action time of the protection device, and compare the actual action behavior and actual action time with the expected action results;

[0038] The verification is considered successful when the deviation between the actual action and the actual action time and the expected action result is within the preset threshold range; otherwise, the deviation item is marked and a verification conclusion report containing verification path tracing information is output.

[0039] Secondly, this application also provides a verification system for relay protection, comprising:

[0040] The acquisition module is used to acquire the instruction manual text of the relay protection device;

[0041] The analysis module is used to parse the action logic information in the instruction manual text, extract the condition information and corresponding action information from the action logic information, and obtain a set of condition attributes.

[0042] The building module is used to construct a validation decision tree based on a set of conditional attributes. The validation decision tree includes multiple validation paths extending from the root node. Each validation path includes multiple branch nodes and corresponding leaf nodes. Branch nodes map conditional information, and leaf nodes map action information.

[0043] The matching module is used to match the corresponding target verification path in the verification decision tree according to the verification requirements of the relay protection device.

[0044] The verification module is used to verify the operation of the relay protection device by taking the target condition information in the target verification path as the verification condition and the target action information in the leaf node corresponding to the target condition information as the expected result, and obtaining the corresponding verification conclusion report.

[0045] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the relay protection verification method provided in the first aspect.

[0046] Fourthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the relay protection verification method provided in the first aspect.

[0047] The aforementioned verification method, device, computer equipment, and program product for relay protection obtain the instruction manual text of the relay protection device; parse the action logic information in the instruction manual text, and extract the condition information and corresponding action information from the action logic information to obtain a set of condition attributes; based on the set of condition attributes, construct a verification decision tree; wherein the verification decision tree includes multiple verification paths extending from the root node, and each verification path includes multiple branch nodes and corresponding leaf nodes, with branch nodes mapping condition information and leaf nodes mapping action information; according to the verification requirements of the relay protection device, match the corresponding target verification path in the verification decision tree; use the target condition information in the target verification path as the verification condition, and use the target action information in the corresponding leaf node as the expected result to verify the operation of the relay protection device, and obtain the corresponding verification conclusion report. In this way, by automatically parsing the instruction manual text of the relay protection device and constructing the verification decision tree, the objective execution of logic verification is achieved, reducing reliance on manual labor, improving the completeness and efficiency of verification, and facilitating adaptation to relay protection devices from different manufacturers. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a diagram illustrating the application environment of a relay protection verification method in one embodiment.

[0050] Figure 2 This is a schematic diagram of the first process in one embodiment of the verification method for relay protection;

[0051] Figure 3 This is a schematic diagram of the second process in one embodiment of the verification method for relay protection;

[0052] Figure 4 This is a schematic diagram of the third process in one embodiment of the verification method for relay protection;

[0053] Figure 5 This is a schematic diagram of a relay protection verification system in one embodiment;

[0054] Figure 6 This is a schematic diagram of the internal structure of a computer device in one embodiment. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0057] The verification method for relay protection provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on another network server.

[0058] Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, aircraft, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can include virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc.

[0059] Server 104 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0060] In the field of power system relay protection technology, the operating logic of relay protection devices is usually presented in unstructured natural language in manufacturer manuals and regulations, making it impossible for machines to directly read and automate the calculations. Manual testing relies heavily on the professional experience of testers, is highly subjective, and struggles to comprehensively cover all boundary conditions when faced with complex logical combinations, particularly the boundary conditions of blocking logic combinations, which are easily overlooked, resulting in insufficient test coverage. Furthermore, existing methods attempting to describe protection logic using general modeling languages ​​suffer from complex modeling rules and a disconnect from physical load testing, making it difficult to adapt to the differentiated characteristics of relay protection devices from different manufacturers, further limiting the level of automation and applicability of testing and verification.

[0061] For example, during the grid connection testing of a line protection device in a 500kV smart substation, testers verified the operating logic of distance protection stage III according to the instruction manual. The manual described the logic in natural language as follows: "When the measured impedance falls into the operating area, and the directional element determines a positive fault, and there is no TV disconnection alarm or control word activation, the circuit breaker will trip after a delay." However, when manually designing test cases, the testers failed to fully consider the boundary combination of simultaneous TV disconnection and directional element abnormalities. This resulted in the omission of this specific scenario in the physical test, causing the device to fail to properly lock out such faults, posing a potential risk of maloperation. In this scenario, the unstructured nature of the natural language description makes the extraction and association of conditional information highly dependent on manual judgment. The complexity of the locking condition combinations further exacerbates the difficulty of covering boundary conditions, directly affecting the completeness and reliability of the test results.

[0062] If the above problems are not addressed, the testing and verification of relay protection devices will remain reliant on human experience, making it difficult to guarantee the objectivity and completeness of the testing process. Insufficient test coverage may lead to the failure to detect defects in the protection logic in a timely manner, resulting in protection maloperation, failure to operate, or timeouts during actual grid operation, thereby affecting the safe and stable operation of the power system. Furthermore, the differentiated implementations of devices from different manufacturers will further increase the complexity of test adaptation, hindering the standardization and automation of relay protection testing technology, making it difficult for testing efficiency to meet the growing functional complexity requirements of smart grids.

[0063] Based on this, in an exemplary embodiment, such as Figure 2 As shown, a verification method for relay protection is provided, which is then applied to... Figure 1 Taking server 104 as an example, the verification method for this relay protection may include steps 21 to 25, wherein:

[0064] Step 21: Obtain the instruction manual text of the relay protection device.

[0065] Here, a relay protection device is an automated device used in a power system. Its main function is to quickly and accurately detect faults and issue trip commands when a fault occurs in the power system, so as to isolate faulty components, protect power equipment from damage, and maintain the stable operation of the power system.

[0066] The instruction manual can refer to the technical documents accompanying the relay protection device, which details the functions, performance parameters, operating logic, setting values, wiring methods, and operating procedures of the relay protection device. The instruction manual is usually presented in natural language.

[0067] It should be noted that the instruction manual is fundamental to understanding the operating logic of relay protection devices. The content of the printed instruction manual can be manually entered into a computer system, or the printed document can be converted into an electronic image file using a scanning device, and then converted into an editable text format using Optical Character Recognition (OCR). Alternatively, if the instruction manual itself is already an electronic document, it can be directly obtained through file import.

[0068] Step 22: Parse the action logic information in the instruction manual text, and extract the condition information and corresponding action information from the action logic information to obtain the condition attribute set.

[0069] Among them, the action logic information can refer to the inherent rules and judgment process of the relay protection device to perform specific actions under specific conditions. The action logic in the action logic information is usually described in the form of "if...then...", which contains instructions on how the device responds to various changes in electrical quantities and internal states.

[0070] Specifically, condition information can refer to various preconditions for triggering or blocking the operation of a relay protection device. Condition information can be threshold values ​​of electrical quantities (e.g., current or voltage exceeding a certain set value), time delays, the status of internal control words of the device, external signal inputs, etc.

[0071] In addition, action information can refer to the specific operations performed by the relay protection device after certain conditions are met, such as issuing a trip command, initiating reclosing, issuing an alarm signal, or recording an event.

[0072] Furthermore, after determining the condition information and action information in the action logic information, a corresponding set of condition attributes can be generated. The set of condition attributes can refer to the dataset formed after structuring and standardizing the condition information and corresponding action information extracted from the instruction manual text. Each element in the condition information can be used to represent an attribute that can be used to build a logical model and its possible values.

[0073] As one approach, after obtaining the instruction manual, professionals can read the manual and manually identify the statements describing the operating logic of the relay protection device. They can then manually mark the conditional and operational parts of the statements and organize this information into a structured table.

[0074] As another implementation method, a series of keywords and phrases can be predefined, such as "when...", "if...then...", "satisfy the condition", etc., and then a simple text matching algorithm can be used to search for these keywords in the instruction manual text, and condition information and action information can be initially extracted based on the context of the keywords.

[0075] For example, when a statement such as "trip when the current is greater than 10A" is detected, "current greater than 10A" can be extracted as condition information and "trip" as action information.

[0076] Step 23: Construct a validation decision tree based on the set of conditional attributes.

[0077] The verification decision tree includes multiple verification paths extending from the root node. Each verification path includes multiple branch nodes and corresponding leaf nodes. Branch nodes map condition information, and leaf nodes map action information.

[0078] It should be noted that the verification decision tree can be a tree-structured model used to represent the operation logic of a relay protection device. Each internal node (branch node) in the verification decision tree can be used to represent a condition judgment, each branch on the branch node can be used to represent a judgment result, and each leaf node represents a final action.

[0079] The root node is the starting point of the verification decision tree and can be used to represent the starting point of a verification tree that corresponds one-to-one with the current relay protection device. Branch nodes are intermediate nodes in the verification decision tree, excluding the root and leaf nodes; each branch node represents a condition judgment. Leaf nodes are the ending nodes of the verification decision tree; each leaf node maps to the action information that the relay protection device will execute after all condition judgments from the root node to that leaf node are satisfied.

[0080] Furthermore, after constructing the verification decision tree, the verification decision tree can include multiple verification paths extending from the root node. A verification path can refer to a complete path from the root node of the verification decision tree to any leaf node. Each verification path can be used to represent the specific action logic that the relay protection device will execute under a specific combination of conditions.

[0081] A decision tree is a model that visually represents the operational logic of a relay protection device. As an example, the construction process can be done manually by drawing the structure of the decision tree based on a pre-organized set of conditional attributes. Each conditional attribute is treated as an internal node of the tree, and the operational information is ultimately used as a leaf node based on the value of the conditional attribute. For example, for the conditional attribute "whether the current is greater than 10A", a branch node can be set, leading to two branches: "yes" and "no".

[0082] As an alternative, a simple set of rules can be designed, for example, to sequentially use the conditional attributes as branch nodes of the decision tree according to the order in which they appear in the text or a preset priority, and to create branches according to their possible values, until all conditional attributes have been considered or a preset depth has been reached, and finally the corresponding action information is used as the leaf node.

[0083] Step 24: Match the corresponding target verification path in the verification decision tree according to the verification requirements of the relay protection device.

[0084] Verification requirements can refer to the strategies or objectives used when testing and verifying relay protection devices. For example, it may be necessary to cover all possible action logic paths or to prioritize the verification of certain key boundary conditions.

[0085] Here, the target verification path can refer to one or more verification paths selected from the verification decision tree for actual testing based on specific verification requirements.

[0086] For example, if the verification requirement is to test the protection action under a specific fault scenario, a path matching the conditions of that fault scenario can be manually selected from the decision tree as the target verification path. Alternatively, some simple matching rules can be preset, such as selecting a path containing specific conditional attributes (e.g., "overcurrent"), or selecting the shortest path as the target verification path.

[0087] Step 25: Use the target condition information in the target verification path as the verification condition, and use the target action information in the leaf node corresponding to the target condition information as the expected result to verify the operation of the relay protection device and obtain the corresponding verification conclusion report.

[0088] The target condition information can refer to all the condition judgments contained in the target verification path. The target condition information will be set as the input of the relay protection device to simulate a specific operating scenario.

[0089] Here, target action information can refer to the action mapped by the leaf node of the target verification path. Target action information is the response that the relay protection device should have under the target condition information.

[0090] In addition, the expected result can refer to the target action information that the relay protection device should theoretically perform under given target conditions.

[0091] Furthermore, the verification conclusion report can be a document generated after the verification process of the relay protection device is completed. It can record the test conditions, expected results, actual results, and the conclusion of whether the verification is passed, and may include any deviations found.

[0092] One approach is to manually set the corresponding test parameters on the relay protection tester based on the target conditions. For example, specific current and voltage values ​​can be input, along with appropriate delays. Then, the tester is started to execute the test, and the indicator lights and display screen of the relay protection device are observed manually, or the device's action signals are acquired using simple signal acquisition equipment. The observed actual actions are compared with the expected results, and the comparison results are manually recorded to generate a verification conclusion report.

[0093] For example, if the expected result is "tripping", and the device actually trips, then it is recorded as verification passed.

[0094] In the traditional relay protection verification process, testers need to manually read the instruction manual of the relay protection device and design test cases based on their personal experience and understanding. This method is not only inefficient, but also prone to overlooking key boundary conditions or specific logic combinations when faced with complex action logic, resulting in insufficient test coverage and thus failing to fully guarantee the reliability of the relay protection device.

[0095] In the above implementation, by automatically parsing the instruction manual text, the unstructured natural language description is transformed into a set of machine-computable conditional attributes, and further constructed into a formalized verification decision tree, thus realizing the machine computability and visualization of the protection logic. Each verification path in the decision tree clearly corresponds to a specific action logic, which allows the test case generation process to shift from being driven by human experience to being model-driven. By using the conditional information in the target verification path as verification conditions and the action information in the leaf nodes as expected results, physical increment testing can be automatically executed, and the actual actions can be compared with the expected results. This improves the objectivity and consistency of test judgment, avoids the subjective bias that may be introduced by human judgment, and can accurately determine whether the device trips within the expected time. It also automatically generates a verification conclusion report containing traceability information without the need for manual observation and recording. Through an automated and standardized verification process, the problems of strong test subjectivity, easy omission of boundary conditions, and low test efficiency in existing technologies are solved.

[0096] In some of the above embodiments, this application proposes a verification method for relay protection. After obtaining the instruction manual of the relay protection device, it is necessary to parse the action logic information in the instruction manual and extract the condition information and corresponding action information from the action logic information to obtain a set of condition attributes. However, the instruction manual is usually an unstructured natural language description. Directly and accurately extracting the condition information and action information used to construct the verification decision tree from it and structuring it faces problems such as complex semantic understanding, inaccurate information extraction, and difficulty in standardization.

[0097] Based on this, in an exemplary embodiment, such as Figure 3 As shown, step 22 involves parsing the action logic information in the instruction manual text and extracting the conditional information and corresponding action information from the action logic information to obtain the conditional attribute set, which includes steps 221 to 224, wherein:

[0098] Step 221: Identify paragraphs describing action logic in the instruction manual text using natural language processing technology, and extract key information from these paragraphs.

[0099] Among them, key information is used to represent the action logic information of relay protection devices, and natural language processing technologies include named entity recognition or domain-based regular expression matching.

[0100] Here, key information can be automatically identified and located from the unstructured instruction manual of the relay protection device, focusing on the core content related to the relay protection operation logic.

[0101] As one approach, instruction manuals typically contain a large amount of non-action logic descriptions, such as product introductions and installation instructions. A Named Entity Recognition (NER) model can be trained to focus on the effective information in the instruction manual and identify predefined entity types, such as "fault type," "electrical quantity threshold," "action delay," "trip," and "alarm," thereby marking sentences or paragraphs containing these entities as key information.

[0102] As another implementation method, a series of keywords and phrase patterns for the field of relay protection can be predefined and matched using regular expressions. For example, rules can be defined to match sentences containing phrases such as "when..., then..." or "if..., then...", as well as paragraphs containing specific relay protection terms, thereby extracting key information of the action logic.

[0103] Step 222: Extract condition information and corresponding action information from the key information.

[0104] The condition information includes action conditions and interlocking conditions. The action conditions include at least the fault type, electrical quantity threshold, delay time and directional element status. The interlocking conditions include at least the control word status, soft pressure plate status and external abnormal status. The action information includes at least tripping, alarm and reclosing.

[0105] Based on the identified key information, the "condition information" that triggers the action in the relay protection operation logic and the "action information" that is executed after the condition is met can be further extracted. At the same time, the condition information can be classified, clarifying that the condition information can include action conditions and blocking conditions.

[0106] As one approach, dependency parsing techniques from natural language processing can be used to analyze sentence structure within key information, identify subject-verb-object relationships, and thus distinguish between conditional clauses and result clauses. Combined with semantic role labeling, this can further identify words representing specific conditional elements such as "fault type" and "electrical quantity threshold," as well as words representing actions such as "tripping" and "alarm."

[0107] As another implementation method, extraction templates can be preset for different types of condition information and action information. For example, for "fault type", keywords such as "single-phase grounding" and "phase-to-phase short circuit" can be matched; for "electrical quantity threshold", modes such as "current greater than 5 amps" and "voltage less than 90 volts" can be matched.

[0108] Step 223: Match and associate the extracted conditional information with the corresponding action information to obtain the conditional action mapping dataset.

[0109] Among them, the conditional action mapping dataset can be used to represent the explicit correspondence between conditions and actions.

[0110] In relay protection logic, specific combinations of conditions can trigger specific actions. When extracting conditional and action information, their positions in the original text and their corresponding syntactic structures can be recorded simultaneously. If conditional and action information appear in the same conditional or causal sentence, or are semantically closely related, they can be matched and associated. For example, by analyzing a sentence structure like "When condition A is met, execute action B," A and B can be associated, resulting in a conditional-action mapping database that includes both conditional and action information.

[0111] Step 224: Construct a condition attribute library based on the condition action mapping dataset to structure the condition information and action information in the condition action mapping dataset and obtain a set of condition attributes.

[0112] Among them, the set of conditional attributes can be transformed from the associated conditional and action information extracted from unstructured text into standardized, machine-understandable structured data.

[0113] Here, the conditional attribute library provides a unified framework that allows similar conditions and actions in different specifications to be normalized.

[0114] As an example, structuring the condition and action information in the condition-action mapping dataset can be achieved by pre-constructing an ontology in the relay protection field, which defines the concepts and attributes of various conditions (such as fault types and electrical quantity thresholds) and actions (such as tripping and alarms). Then, the unstructured text descriptions in the condition-action mapping dataset are mapped to the corresponding concepts and attributes in the ontology using semantic mapping techniques, thereby achieving structuring and obtaining the corresponding set of condition attributes.

[0115] As another example, standard data patterns and data types can be defined for different types of conditional and action information. By filling the extracted information into these predefined patterns and performing type conversion and standardization, a structured set of conditional attributes can be formed.

[0116] Through the above technical solution, this application effectively solves the problems of low efficiency, easy omissions, and errors in manually extracting action logic information from unstructured instruction manual texts in traditional relay protection verification. By introducing natural language processing technology, the automatic identification and key information extraction of action logic from instruction manual texts are achieved, significantly improving the accuracy and efficiency of information acquisition. Furthermore, the condition information is meticulously classified (action conditions and blocking conditions), and the specific condition elements and action types are clearly defined, making the extracted information more standardized and complete. Finally, through the construction of condition action mapping and condition attribute library, unstructured text information is transformed into a standardized, machine-readable set of condition attributes, providing a high-quality, structured data foundation for the subsequent automated construction of verification decision trees. This not only significantly reduces the workload of manual analysis and data preparation but also ensures the accuracy and consistency of verification logic, thereby comprehensively improving the automation level and reliability of relay protection verification.

[0117] In one embodiment, the step of constructing a standard conditional attribute library based on the conditional action mapping dataset to structure the conditional and action information in the dataset and obtain a set of conditional attributes includes:

[0118] (1) Establish a condition attribute library.

[0119] The condition attribute library includes at least the following: a fault type attribute library (discrete type, including single-phase grounding, phase-to-phase short circuit, and three-phase short circuit); an electrical quantity threshold attribute library (continuous type, including overcurrent setting, impedance setting, and zero-sequence voltage setting); a delay time attribute library (numerical type, including action delay); a control word status attribute library (Boolean type, including enabled and disabled); and a soft pressure plate status attribute library (Boolean type, including enabled and disabled).

[0120] It should be noted that the condition attribute library can be a standardized data structure used to store and manage condition information of relay protection devices. The condition attribute library aims to provide a unified classification, naming and description specification for various condition information, thereby ensuring data consistency and processability.

[0121] You can build a conditional attribute library by designing a relational database table structure that includes fields for defining different conditional attributes; or by building an XML or JSON-based configuration file that uses predefined tags and structures to describe various conditional attributes.

[0122] The fault type attribute library can be a specific component of the condition attribute library, used to define and manage various fault types that relay protection devices may encounter. The fault type attribute library can be discrete, meaning that it contains a limited and enumerable number of fault types, such as single-phase grounding, phase-to-phase short circuit, and three-phase short circuit. The fault type attribute library can be constructed by creating enumeration fields in the database or defining a fixed list of fault type attributes in the configuration file.

[0123] Furthermore, the electrical quantity threshold attribute library can be another important component of the condition attribute library. It is used to define and manage threshold conditions related to electrical quantities. The electrical quantity threshold attribute library can be continuous, meaning that the thresholds it contains can be any value within a certain range, such as overcurrent settings, impedance settings, and zero-sequence voltage settings. The electrical quantity threshold attribute library can be constructed by defining floating-point fields in the database and attaching range constraints; or by specifying the numeric type and possible value range in a configuration file.

[0124] The delay time attribute library is used to define and manage the delay time conditions required for the operation of relay protection devices. The delay time attribute library can be numeric, meaning that the delay time it contains is a specific numerical value, such as the action delay. Numeric types can be represented by integer or floating-point numbers in the data structure, and the time unit can be specified.

[0125] The control word status attribute library is used to define and manage the status conditions of control words in relay protection devices. The control word status attribute library can be Boolean, meaning its status has only two possibilities, such as active and deactivated. Boolean types can be represented using Boolean type fields or integer values ​​of 0 / 1 in the data structure.

[0126] The soft circuit board status attribute library is used to define and manage the status conditions of soft circuit boards in relay protection devices. The soft circuit board status attribute library is also designed as a Boolean type, with only two states: engaged and disengaged. Its implementation is similar to the control word status attribute library, and it can use Boolean type fields or integer values ​​of 0 or 1.

[0127] (2) Classify the condition information in the condition action mapping dataset according to the category of the condition attribute library, and label the condition attribute library with data type, physical quantity type, unit, variable symbol and manufacturer mapping information to obtain the condition attribute set.

[0128] Based on the classification criteria defined in the pre-established condition attribute library, the original condition information extracted from the instruction manual text is identified and matched, and then assigned to the corresponding attribute category.

[0129] One approach is to use a rule engine based on keyword matching. For example, if words related to "fault" are identified, they can be categorized into the fault type attribute library. Alternatively, a machine learning classification model can be used to perform semantic analysis on the text description of the conditional information and automatically categorize it.

[0130] Labeling the condition attribute library with data types, physical quantity types, units, variable symbols, and manufacturer mapping information adds additional metadata to each condition attribute, providing a richer and more accurate description. Data types specify whether the attribute is discrete, continuous, numerical, or Boolean; physical quantity types indicate whether the attribute represents physical quantities such as current, voltage, or time; units clarify the measurement standard for numerical or continuous attributes; variable symbols provide a unified identifier used in the program or model; and manufacturer mapping information associates general attributes with specific parameter names or codes of equipment from particular manufacturers.

[0131] In the aforementioned technical solution, the raw, unstructured conditional information extracted from the instruction manual of the relay protection device is transformed into a structured, standardized set of conditional attributes rich in metadata by establishing a standardized conditional attribute library, classifying and annotating it in detail. This effectively solves the problems of inconsistent raw data formats and ambiguous semantics, greatly improving the consistency and processability of conditional information. The structured set of conditional attributes provides high-quality, standardized input for subsequent construction of verification decision trees, thereby significantly improving the accuracy and efficiency of decision tree construction. Furthermore, by annotating data types, physical quantity types, units, variable symbols, and manufacturer mapping information, this method can better adapt to the verification needs of relay protection devices from different manufacturers and of different models, enhancing the method's versatility and automation, reducing the probability of manual intervention and errors, and ultimately improving the reliability and efficiency of relay protection verification.

[0132] In some of the above implementations, a method for constructing a verification decision tree based on a set of conditional attributes to verify relay protection devices has been proposed. However, in practice, how to systematically, efficiently, and accurately construct a verification decision tree from a complex set of conditional attributes that can effectively reflect the operating logic of the relay protection device, so as to ensure the comprehensiveness and logic of the verification path, is a technical problem that needs to be solved.

[0133] Based on this, such as Figure 4 As shown, step 23, constructing the validation decision tree based on the set of conditional attributes, includes steps 231 to 233, wherein:

[0134] Step 221: Using the conditional attributes in the conditional attribute set as feature dimensions and the corresponding action information as classification labels, construct a labeled multidimensional dataset.

[0135] Among them, the multidimensional dataset can be a structured data form that is transformed from the set of conditional attributes and corresponding action information parsed and extracted from the instruction manual text.

[0136] Specifically, each conditional attribute can be considered a feature dimension of the dataset, while the action information corresponding to each combination of conditional attributes can be considered a classification label for the dataset, thus transforming unstructured logical descriptions into computable mathematical models. For example, the data can be organized into a table, where each row represents a specific instance of action logic, and the columns correspond to different conditional attributes (such as fault type, electrical quantity threshold, etc.) and the final action information (such as tripping, alarm).

[0137] Step 222: Traverse the conditional attributes in the multidimensional dataset and calculate the information gain of multiple conditional attributes.

[0138] Information gain is a key metric used in decision tree algorithms to evaluate the importance of features. It quantifies the degree to which the uncertainty (entropy) of a dataset is reduced after a certain conditional attribute is given.

[0139] It's important to note that calculating the information gain of each conditional attribute measures its ability to distinguish between different action information. A higher information gain indicates greater effectiveness of the conditional attribute in classifying action information. Information gain can be used to ensure the selection of the most discriminative conditional attribute as the split point, thereby constructing a more structurally sound and efficient decision tree.

[0140] In one implementation, the step of traversing the conditional attributes in the dataset and calculating the information gain of each conditional attribute includes:

[0141] (1) Calculate the information entropy of the multidimensional dataset based on the frequency of occurrence of various action information in the multidimensional dataset.

[0142] Information entropy is used to represent the uncertainty measure of a multi-dimensional dataset.

[0143] Information entropy can be used mathematically to quantify the degree of uncertainty or disorder of action information in the multidimensional dataset, and is used to assess the purity of the action information distribution before considering any conditional attributes.

[0144] As one implementation method, it can be calculated using the Shannon entropy formula. For example, for a set containing multiple action information, its information entropy can be obtained by taking the negative logarithm of the probability of each action information occurring and summing the results.

[0145] (2) Divide the multidimensional dataset into multiple dimensional subsets according to the value of the conditional attribute, calculate the information entropy of the multiple dimensional subsets, and sum them up by weight according to the number of samples in each dimensional subset to obtain the conditional entropy of the conditional attribute.

[0146] In this context, a dimensional subset can be a subset obtained by logically dividing the original multi-dimensional dataset according to the different values ​​of the conditional attribute to be evaluated. For example, if a conditional attribute is "fault type", its values ​​may include "single-phase grounding", "phase-to-phase short circuit" and "three-phase short circuit", then the multi-dimensional dataset will be divided into three corresponding subsets, each containing samples with that specific "fault type" value.

[0147] It should be noted that different strategies can be used to divide multi-dimensional data into multiple dimensional subsets depending on the type of conditional attribute (such as discrete or continuous). For example, for discrete attributes, each value corresponds to a subset; for continuous attributes, a threshold can be set to perform binary or multi-subject division to obtain the corresponding multiple dimensional subsets.

[0148] For each dimensional subset obtained from the above division, calculate the information entropy used to represent the uncertainty of the distribution of internal action information.

[0149] Conditional entropy is the sum of the products of the information entropy of each dimensional subset and its proportion in the original multidimensional dataset (i.e., the proportion of the number of samples). It can be used to reflect how much the uncertainty of action information is reduced on average after the value of a certain conditional attribute is known.

[0150] (3) Subtract the conditional entropy of the conditional attribute from the information entropy of the dataset to obtain the information gain of the conditional attribute.

[0151] After partitioning based on a certain conditional attribute, the greater the reduction in uncertainty of the multidimensional dataset and the greater the information gain, the more effective the conditional attribute is for predicting action information, that is, the more classification information the attribute contains.

[0152] By precisely calculating information entropy, conditional entropy, and information gain, the contribution of each conditional attribute to the classification of action information was quantified.

[0153] When constructing the validation decision tree, the information entropy of the entire multi-dimensional dataset is first calculated to measure the initial uncertainty. Then, for each conditional attribute, the dataset is divided into multiple dimensional subsets based on its value, and the information entropy of each subset is calculated separately. The conditional entropy of the conditional attribute is obtained by weighted summation of these subset entropies, representing the uncertainty of action information even when the conditional attribute is known. Finally, the information gain of the conditional attribute is obtained by subtracting the conditional entropy from the initial information entropy. A larger information gain value indicates a stronger discriminative ability of the conditional attribute in classifying action information. This ensures that when constructing the validation decision tree, conditional attributes with the highest discriminative power for action information are prioritized as split nodes, resulting in a more optimized decision tree structure that more accurately reflects the action logic of the relay protection device. This quantitative evaluation method makes the decision tree construction process more scientific and efficient, avoiding errors caused by subjective judgment.

[0154] Step 223: Select the conditional attribute with the largest information gain as the current split node, until the action information of each conditional attribute in the multidimensional dataset is uniquely determined or the conditional attributes are exhausted, thus forming a validation decision tree.

[0155] In the construction of a decision tree, each internal node represents a judgment on a conditional attribute. The conditional attribute with the highest information gain is selected as the current split node. On the current dataset, this attribute minimizes uncertainty and most effectively distinguishes different action information, aiming to make locally optimal choices at each step to achieve a globally optimal decision tree structure. In this way, the decision tree prioritizes the conditional attributes that have the greatest impact on action logic, making the verification path logic clearer and more efficient.

[0156] This process continues until the action information of each conditional attribute in the multidimensional dataset is uniquely determined or the conditional attributes are exhausted, ultimately forming the verification decision tree.

[0157] When all instances contained in a node belong to the same action information category (i.e., the action information is uniquely determined), the node becomes a leaf node and no further splitting is needed, because it can already clearly provide the verification result.

[0158] Another termination case is when all available conditional attributes have been used as split nodes, but the action information under some nodes is still not unique (i.e., conditional attributes are exhausted). In this case, the node can also be regarded as a leaf node, and its action information usually takes the majority category of the samples under the node. The termination condition ensures the integrity and finiteness of the decision tree, avoids infinite recursion, and guarantees that each path from the root node to the leaf node represents a complete verification logic.

[0159] For example, the following proposes a method for constructing a validation decision tree. Specifically, action information is used as the leaf nodes of the validation decision tree, and condition information is used as the internal nodes (branch nodes). The ID3 variant algorithm is used to automatically construct the decision tree. The ID3 variant algorithm incorporates threshold discretization processing for continuous attributes such as electrical quantities, and calculates the information gain of each condition attribute by traversing all combinations of action conditions.

[0160] Here, all action condition combinations can be combinations of all condition information and corresponding action information.

[0161] Assume the current dataset is The total number of samples is , No. The probability of the action result is The information entropy of the dataset is:

[0162]

[0163] For a certain conditional attribute 'a' (such as "PT disconnection status"), if based on this attribute... Divided into If there are subsets, then the information gain of this attribute is:

[0164]

[0165] Its basic execution logic is to prioritize selecting the conditional attribute with the largest information gain (i.e., the one that best distinguishes whether to take an action or the different results of actions) for node splitting. This process is executed recursively until the end of each branch can uniquely determine an action result (or no action), forming the final decision tree.

[0166] Through the above technical solution, this application can automatically construct a clear and logically rigorous verification decision tree from the instruction manual of a relay protection device in a data-driven and intelligent manner. The information gain-based decision tree construction method ensures that the generated verification path accurately reflects the actual operating logic of the relay protection device, avoiding omissions, errors, or inefficiencies that may occur with manual decision tree construction. Therefore, it improves the automation level and reliability of relay protection verification work, making the verification process of relay protection device operation more efficient, comprehensive, and accurate, thereby effectively ensuring the safe and stable operation of the power system.

[0167] In some of the above implementations, a method for matching target verification paths by constructing a verification decision tree is proposed. However, in actual relay protection verification processes, verification requirements may vary. For example, it may be necessary to fully cover all possible action logic, or to focus on certain key or error-prone combinations of conditions. If only a single path matching method is used, it may not be able to flexibly and efficiently meet the verification objectives under different scenarios, resulting in low verification efficiency or insufficient verification coverage.

[0168] Based on this, in an exemplary embodiment, the verification requirements include a full path coverage strategy, a boundary condition priority strategy, or a specified path strategy. Step 24, matching the corresponding target verification path in the verification decision tree according to the verification requirements of the relay protection device may include the following methods.

[0169] It is important to note that verification requirements refer to the verification direction and scope determined during the relay protection verification process, based on different verification objectives and focuses. These strategies are the core principles guiding the selection of verification paths.

[0170] Method 1: When using the full path coverage strategy, traverse all paths from the root node to each leaf node of the verification decision tree to generate the target verification path set.

[0171] The full-path coverage strategy aims to ensure that all possible operational logic paths of the relay protection device are verified. This ensures comprehensive verification and helps identify potential logical errors or omissions. For example, depth-first search or breadth-first search algorithms can be used to traverse all branches of the decision tree.

[0172] When using the full path coverage strategy, the library traverses all paths from the root node to each leaf node of the verification decision tree to generate the target verification path set.

[0173] It's important to note that the purpose of the full-path coverage strategy is to ensure the comprehensiveness of verification. Traversing the verification decision tree means systematically exploring every branch of the tree, starting from the root node and proceeding downwards along each branch until reaching each leaf node. Each complete path from the root node to a leaf node represents a complete action logic of the relay protection device under a specific combination of conditions. By generating a target verification path set, it can be ensured that all possible combinations of action logic are included in the verification scope, thereby improving the thoroughness of verification.

[0174] For example, a recursive or iterative algorithm can be used to traverse the decision tree, storing each complete path as an independent verification path.

[0175] Alternatively, when using a boundary condition priority strategy, the path containing the combination of blocking conditions and the boundary value of electrical quantity thresholds is selected as the target verification path.

[0176] Boundary condition priority strategies focus on verifying the behavior of relay protection devices under critical conditions. Boundary conditions are often where problems most likely to occur; for example, the upper and lower limits of electrical quantity thresholds, and combinations of various blocking conditions. This helps to identify potential defects in the device under extreme or abnormal operating conditions. For example, a set of boundary value rules can be preset, prioritizing the matching of paths containing these rules.

[0177] When employing a boundary condition priority strategy, paths containing both blocking condition combinations and electrical quantity threshold boundary values ​​can be prioritized as target verification paths. This strategy aims to focus on the critical and sensitive areas of relay protection devices. A blocking condition combination refers to the simultaneous fulfillment or non-fulfillment of multiple blocking conditions (such as control word status or soft-plate status) in a specific manner; these combinations often determine the activation or deactivation of protection functions.

[0178] Electrical quantity threshold boundary values ​​refer to the conditions where electrical quantities such as current, voltage, and impedance are near or at their set values ​​or limits. These are the critical points for protection action criteria. Prioritizing paths that include these characteristics can effectively test the performance and reliability of the device at these critical points.

[0179] For example, when traversing the decision tree, paths containing predefined combinations of blocking conditions or electrical quantity threshold boundary values ​​can be given higher priority, or paths that meet these conditions can be selected from all paths.

[0180] Alternatively, method three: When using a specified path strategy, the corresponding verification path is located based on the combination of input condition attributes as the target verification path.

[0181] Specifying a path allows users to precisely select one or more verification paths based on specific testing objectives or known failure scenarios. This provides high flexibility and targeting, making it suitable for quickly reproducing problems or verifying specific functions. For example, users can directly specify a path by entering a series of conditional attribute values.

[0182] When using a specified path strategy, the corresponding verification path can be located as the target verification path based on the combination of input conditional attributes. The specified path strategy provides high flexibility and accuracy. Users can input a series of conditional attribute combinations according to specific verification requirements, such as for a known fault type, a specific delay setting, or a control word state.

[0183] Based on the conditional attributes of these inputs, matching and positioning are performed in the constructed verification decision tree to find the unique or most suitable verification path for that combination, making the verification process highly customizable and quickly focusing on specific verification scenarios.

[0184] For example, the validation path corresponding to the combination of input condition attributes can be quickly located by performing condition matching search in a decision tree or by using hash mapping.

[0185] In the above method, the instruction manual of the relay protection device is obtained, the action logic information is parsed, and condition and action information are extracted to construct a verification decision tree. Based on this, to more flexibly and efficiently meet diverse relay protection verification needs, this application further introduces multiple verification strategies. When verification is required, instead of simply matching arbitrary paths, a target verification path is intelligently selected or generated according to preset verification requirements. Specifically, if the verification target is comprehensiveness, a full-path coverage strategy is adopted, systematically traversing all paths from the root node to the leaf node in the verification decision tree to ensure that every possible combination of action logic is considered, thereby generating a comprehensive set of target verification paths. If the verification target targets the device's weak points or critical operating conditions, a boundary condition priority strategy is adopted, intelligently identifying and prioritizing paths containing specific blocking condition combinations or electrical quantity threshold boundary values. These paths represent the device's behavior under critical or abnormal conditions and are the focus of verification. When the verification target targets a specific scenario or known problem, a specified path strategy is adopted, allowing users to directly input a series of condition attribute combinations, and the corresponding verification path is accurately located in the verification decision tree based on these inputs. This allows the verification process to be customized according to actual needs, avoiding unnecessary verification work, improving the relevance and efficiency of verification, and ensuring verification coverage of key scenarios, thereby effectively solving the limitations of the single path matching method when facing diverse verification needs.

[0186] The above technical solution enables flexible selection of different verification strategies based on the actual verification needs of relay protection devices, thereby efficiently matching the target verification path that best meets the current verification objective within the constructed verification decision tree. Whether comprehensive logic coverage is required, the behavior of the device under critical conditions needs to be focused on, or rapid verification is needed for specific scenarios, this application can provide corresponding strategy support. This significantly improves the targeting, flexibility, and efficiency of relay protection verification work, avoids unnecessary duplicate verification, ensures coverage of key verification points, and thus improves the quality and reliability of relay protection device verification.

[0187] In some embodiments described above in this application, a method for verifying relay protection devices by constructing a verification decision tree and matching a target verification path is proposed. However, in the actual verification process, how to accurately configure the test environment of the relay protection device based on the target condition information extracted from the decision tree, execute physical increments, and accurately determine whether the actual action behavior of the relay protection device meets expectations is a key challenge to ensure the effectiveness and reliability of the verification.

[0188] Based on this, in an exemplary embodiment, step 25, using the target condition information in the target verification path as the verification condition and the target action information in the leaf node corresponding to the target condition information as the expected result, verifies the operation of the relay protection device and obtains the corresponding verification conclusion report, including steps 251 to 256, wherein:

[0189] Step 251: If the condition attribute in the target verification path is Boolean, then set the state of the corresponding control word and soft pressure plate of the relay protection device.

[0190] Boolean condition attributes refer to conditions that have only two discrete states (e.g., "on" or "off", "enabled" or "disabled"), such as control words or soft pressure plates inside relay protection devices.

[0191] As one implementation method, the state setting can be done manually through the local human-machine interface of the relay protection device, or by sending remote control commands through its communication interface (such as IEC 61850, Modbus and other protocols) to automatically set the corresponding control word or soft pressure plate to the specific state required by the target verification path.

[0192] Step 252: If the condition attribute in the target verification path is continuous, then set the increment parameters of the relay protection tester.

[0193] The parameters to be added include at least the magnitude and phase angle of the fault phase voltage and the fault phase current.

[0194] Continuous condition attributes refer to conditions that can take values ​​continuously within a certain range, such as various electrical quantity thresholds (e.g., overcurrent setting, impedance setting, zero-sequence voltage setting).

[0195] Setting the parameters of a relay protection tester can refer to configuring the tester's analog output function so that it can generate voltage and current signals that conform to specific fault scenarios. The parameters typically include the amplitude and phase angle of the fault phase voltage, the amplitude and phase angle of the fault phase current, and the corresponding parameters of the non-fault phases, in order to accurately simulate various fault types and severity in the power system.

[0196] Another approach is to pre-set a series of increment sequences through the tester's programming interface or dedicated software, enabling it to automatically output according to predetermined fault scenarios, thereby simulating complex fault evolution processes.

[0197] Step 253: Set timing parameters.

[0198] The timing parameters include at least the pre-fault state holding time and the fault state duration.

[0199] Here, timing parameters are used to define the duration of different stages during the test. The pre-fault state holding time refers to the time the relay protection device is in normal operating condition before the simulated fault occurs, ensuring the device is in a stable and normal working state before the fault occurs. The fault state duration refers to the duration of the simulated fault state, ensuring the relay protection device has sufficient time to detect the fault, make a judgment, and issue an action signal.

[0200] In addition, fault recovery time and multiple fault intervals can be set to simulate more complex operating conditions.

[0201] Step 254: Control the relay protection tester to perform physical addition according to the addition parameters, pre-state and timing parameters.

[0202] Controlling the relay protection tester to perform physical input refers to activating the relay protection tester so that it outputs simulated voltage and current signals to the relay protection device according to pre-set electrical quantity parameters (input parameters), the initial state of the relay protection device (pre-set state, such as control word and soft pressure plate settings), and time series parameters (timing parameters). This simulates the fault occurrence process in an actual power system, enabling the relay protection device to respond in its real operating environment.

[0203] Step 255: Monitor the real-time action signals, operating data signals, and switch on / off signals fed back by the relay protection device, determine and calculate the actual action behavior and actual action time of the protection device, and compare the actual action behavior and actual action time with the expected action result.

[0204] Real-time action signals typically refer to binary signals such as tripping and alarms issued by the relay protection device. Operating data signals refer to internal measurements and status variables of the relay protection device, which can be obtained through communication interfaces (such as IEC 61850, Modbus, etc.). Switch on / off signals refer to the on / off status signals of circuit breakers and other switching equipment controlled by the relay protection device.

[0205] By collecting and analyzing these signals, it is possible to accurately determine whether the relay protection device has operated (actual operating behavior) and the time interval from the occurrence of the fault to the issuance of the operating signal (actual operating time).

[0206] Step 256: When the deviation between the actual action and the actual action time and the expected action result is within the preset threshold range, the verification is deemed successful; otherwise, the deviation item is marked and a verification conclusion report containing verification path tracing information is output.

[0207] The preset threshold can refer to the allowable error range, such as the allowable millisecond-level deviation in action time.

[0208] Furthermore, the actual measured actions and times are compared with the expected actions (such as tripping and alarms) and expected times (calculated based on protection settings) in the target verification path. If the deviation between the actual actions and times and the expected results is within a preset threshold range, the verification path is considered successful. Conversely, if the deviation exceeds the range, non-compliance is automatically recorded, including the actual value, expected value, and deviation, and a detailed verification conclusion report is generated. This report includes not only test conditions, expected results, actual results, and deviations, but also traceability information for the verification path, facilitating subsequent analysis and problem localization.

[0209] Through the above technical solution, this application can transform the abstract action logic in the instruction manual of a relay protection device into executable and quantifiable physical verification steps. This solution ensures the accuracy and repeatability of the relay protection device testing environment by accurately distinguishing between Boolean and continuous condition attributes and configuring them accordingly. Simultaneously, the introduction of timing parameters and physical increment control makes the fault simulation process more realistic, comprehensively reflecting the device's response under dynamic operating conditions. Furthermore, real-time monitoring of multiple signals and automated comparison with expected results greatly improves verification efficiency and the objectivity of the results, avoiding the subjectivity of human judgment. When deviations occur, the system can automatically mark and output a report containing traceability information, which is crucial for quickly locating problems and analyzing fault causes, thereby significantly improving the automation level, accuracy, and reliability of relay protection device verification.

[0210] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0211] Based on the same inventive concept, this application also provides a relay protection verification system for implementing the aforementioned relay protection verification method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations of the one or more relay protection verification system embodiments provided below can be found in the limitations of the relay protection verification method described above, and will not be repeated here.

[0212] In one exemplary embodiment, such as Figure 5 As shown, a verification system for relay protection is provided, including: an acquisition module, an analysis module, a construction module, a matching module, and a verification module, wherein:

[0213] The acquisition module is used to acquire the instruction manual text of the relay protection device;

[0214] The analysis module is used to parse the action logic information in the instruction manual text, extract the condition information and corresponding action information from the action logic information, and obtain a set of condition attributes.

[0215] The building module is used to construct a validation decision tree based on a set of conditional attributes. The validation decision tree includes multiple validation paths extending from the root node. Each validation path includes multiple branch nodes and corresponding leaf nodes. Branch nodes map conditional information, and leaf nodes map action information.

[0216] The matching module is used to match the corresponding target verification path in the verification decision tree according to the verification requirements of the relay protection device.

[0217] The verification module is used to verify the operation of the relay protection device by taking the target condition information in the target verification path as the verification condition and the target action information in the leaf node corresponding to the target condition information as the expected result, and obtaining the corresponding verification conclusion report.

[0218] Each module in the aforementioned relay protection verification system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0219] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores runtime and verification data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a relay protection verification method.

[0220] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0221] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of a relay protection verification method.

[0222] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program being executed by a processor to implement the steps of a verification method for relay protection.

[0223] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of a verification method for relay protection.

[0224] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0225] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0226] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A verification method for relay protection, characterized in that, The method includes: Obtain the instruction manual for the relay protection device; The action logic information in the instruction manual text is parsed, and the condition information and corresponding action information in the action logic information are extracted to obtain a set of condition attributes; Based on the set of conditional attributes, a verification decision tree is constructed; wherein, the verification decision tree includes multiple verification paths extending from the root node, each verification path includes multiple branch nodes and corresponding leaf nodes, the branch nodes map the conditional information, and the leaf nodes map the action information; Based on the verification requirements of the relay protection device, the corresponding target verification path is matched in the verification decision tree; The target condition information in the target verification path is used as the verification condition, and the target action information in the leaf node corresponding to the target condition information is used as the expected result to verify the operation of the relay protection device and obtain the corresponding verification conclusion report.

2. The method according to claim 1, characterized in that, The step of parsing the action logic information in the instruction manual text and extracting the condition information and corresponding action information from the action logic information to obtain the condition attribute set includes: The paragraphs describing the action logic in the instruction manual are identified using natural language processing technology, and key information is extracted from the paragraphs; wherein, the key information is used to represent the action logic information of the relay protection device, and the natural language processing technology includes named entity recognition or regular expression matching based on domain rules; Extract condition information and corresponding action information from the key information; wherein, the condition information includes action conditions and interlocking conditions, the action conditions include at least fault type, electrical quantity threshold, delay time and directional element status, the interlocking conditions include at least control word status, soft pressure plate status and external abnormal status, and the action information includes at least tripping, alarm and reclosing. The extracted conditional information is matched and associated with the corresponding action information to obtain a conditional action mapping dataset; A condition attribute library is constructed based on the condition action mapping dataset to structure the condition information and action information in the condition action mapping dataset, thereby obtaining the condition attribute set.

3. The method according to claim 2, characterized in that, The step of constructing a standard conditional attribute library based on the conditional action mapping dataset, and structuring the conditional information and action information in the conditional action mapping dataset to obtain the conditional attribute set, includes: Establish a condition attribute library; wherein the condition attribute library includes at least: a fault type attribute library, which is discrete and includes single-phase grounding, phase-to-phase short circuit, and three-phase short circuit; an electrical quantity threshold attribute library, which is continuous and includes overcurrent setting, impedance setting, and zero-sequence voltage setting; a delay time attribute library, which is numerical and includes action delay; a control word status attribute library, which is Boolean and includes "on" and "off"; and a soft pressure plate status attribute library, which is Boolean and includes "on" and "off". The condition information in the condition action mapping dataset is categorized according to the category of the condition attribute library, and the condition attribute library is labeled with data type, physical quantity type, unit, variable symbol and manufacturer mapping information to obtain the condition attribute set.

4. The method according to claim 1, characterized in that, The step of constructing the verification decision tree based on the set of conditional attributes includes: Using the conditional attributes in the set of conditional attributes as feature dimensions and the corresponding action information as classification labels, a labeled multidimensional dataset is constructed. Iterate through the conditional attributes in the multi-dimensional dataset and calculate the information gain of multiple conditional attributes; The conditional attribute with the largest information gain is selected as the current split node, and the process continues until the action information of each conditional attribute in the multidimensional dataset is uniquely determined or the conditional attributes are exhausted, thus forming the verification decision tree.

5. The method according to claim 4, characterized in that, The step of traversing the conditional attributes in the dataset and calculating the information gain of each conditional attribute includes: Based on the frequency of occurrence of various action information in the multi-dimensional dataset, the information entropy of the multi-dimensional dataset is calculated; wherein, the information entropy is used to represent the uncertainty measure of the multi-dimensional dataset; The multidimensional dataset is divided into multiple dimensional subsets based on the value of the conditional attribute. The information entropy of the multiple dimensional subsets is calculated, and the sum is obtained by weighting the sum according to the number of samples in each dimensional subset to obtain the conditional entropy of the conditional attribute. The information gain of the conditional attribute is obtained by subtracting the conditional entropy of the conditional attribute from the information entropy of the dataset.

6. The method according to claim 1, characterized in that, The verification requirements include a full-path coverage strategy, a boundary condition priority strategy, or a specified path strategy. The step of matching the corresponding target verification path in the verification decision tree according to the verification requirements of the relay protection device includes: When using a full path coverage strategy, all paths from the root node to each leaf node of the verification decision tree are traversed to generate a target verification path set; or When using a boundary condition priority strategy, the path containing both the combination of blocking conditions and electrical quantity threshold boundary values ​​is preferentially selected as the target verification path; or When a specified path strategy is used, the corresponding verification path is located as the target verification path based on the combination of input condition attributes.

7. The method according to claim 1, characterized in that, The steps of using the target condition information in the target verification path as the verification condition and the target action information in the leaf node corresponding to the target condition information as the expected result to verify the operation of the relay protection device and obtain the corresponding verification conclusion report include: If the condition attribute in the target verification path is Boolean, then set the state of the corresponding control word and soft pressure plate of the relay protection device; If the condition attribute in the target verification path is continuous, then the increment parameters of the relay protection tester are set; wherein, the increment parameters include at least the magnitude and phase angle of the fault phase voltage and the fault phase current; Set timing parameters; wherein, the timing parameters include at least the pre-fault state holding time and the fault state duration; The relay protection tester is controlled to perform physical application according to the application parameters, pre-state and timing parameters; The system monitors the real-time action signals, operating data signals, and switch on / off signals fed back by the relay protection device to determine the actual action behavior and actual action time of the calculation protection device, and compares the actual action behavior and actual action time with the expected action result. The verification is deemed successful when the deviation between the actual action and the actual action time and the expected action result is within a preset threshold range; otherwise, the deviation item is marked and a verification conclusion report containing verification path tracing information is output.

8. A verification system for relay protection, characterized in that, The device includes: The acquisition module is used to acquire the instruction manual text of the relay protection device; The analysis module is used to parse the action logic information in the instruction manual text, and extract the condition information and corresponding action information from the action logic information to obtain a set of condition attributes. A construction module is used to construct a verification decision tree based on the set of conditional attributes; wherein the verification decision tree includes multiple verification paths extending from the root node, each verification path includes multiple branch nodes and corresponding leaf nodes, the branch nodes map the conditional information, and the leaf nodes map the action information; The matching module is used to match the corresponding target verification path in the verification decision tree according to the verification requirements of the relay protection device. The verification module is used to verify the operation of the relay protection device by taking the target condition information in the target verification path as the verification condition and the target action information in the leaf node corresponding to the target condition information as the expected result, and to obtain the corresponding verification conclusion report.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images