A false operation prevention locking system intelligent diagnosis method for a new energy station

Abstract

This invention relates to the field of anti-misoperation interlocking systems and provides an intelligent diagnostic method for anti-misoperation interlocking systems in new energy power plants. The method includes: constructing and dynamically updating a topology model representing the real-time electrical connections of the power plant by real-time access and fusion of multi-source state information from the power grid data acquisition system, relay protection devices, intelligent grounding wires, and environmental sensors; based on this, using a knowledge graph reasoning model to traverse and detect conflicts in a static anti-misoperation rule base to diagnose logical vulnerabilities; combining historical operational data with a machine learning model to identify potential risk patterns and rule blind spots to improve diagnostic results; analyzing the time-series data of locks and sensors using a predictive model to identify hardware performance degradation trends and fault risks; and finally, synthesizing all the above diagnostic and identification results to generate a structured report. This effectively improves the safety level and adaptive reliability of new energy power plants in complex operating environments.

Description

Technical Field

[0001] This invention relates to the field of anti-misoperation interlocking system technology, and in particular to an intelligent diagnostic method for anti-misoperation interlocking systems in new energy power plants. Background Technology

[0002] In power system substations and new energy power plants, anti-misoperation interlocking systems are key technical equipment for ensuring the safety of switching operations and preventing electrical misoperations. Traditional microcomputer-based anti-misoperation interlocking systems mainly collect the status of auxiliary contacts of primary equipment and combine it with a preset five-prevention logic rule library to perform logic verification before actual operation, thereby achieving interlocking of unauthorized operations. For the management of temporary grounding wires, physical code locks or technologies based on radio frequency identification are usually used for identity verification and location recording.

[0003] However, because the anti-misoperation function relies heavily on a static rule base, the system cannot perceive and verify dynamic electrical topology changes caused by network reconfiguration in real time, and it is particularly difficult to identify hidden safety risks such as backfeeding from distributed power sources. Secondly, the existing system lacks sufficient integration of secondary equipment status monitoring and interlocking logic, resulting in blind spots in anti-misoperation measures. The system lacks predictive diagnostic capabilities for the health status of its own locks, sensors, and other hardware, leading to significant deficiencies in the accuracy, real-time performance, and overall adaptive reliability of the anti-misoperation interlocking system when facing the complex and ever-changing operation and maintenance scenarios of new energy power plants. Summary of the Invention

[0004] This invention provides an intelligent diagnostic method for anti-misoperation interlocking systems in new energy power plants, which addresses the significant deficiencies in the accuracy, real-time performance, and overall adaptive reliability of anti-misoperation interlocking systems when facing complex and ever-changing operation and maintenance scenarios in new energy power plants.

[0005] This invention provides an intelligent diagnostic method for an anti-misoperation interlocking system in new energy power plants, comprising: The system can access the multi-source status information of the power grid data acquisition system, relay protection device, intelligent grounding wire management unit and environmental monitoring sensor associated with the anti-misoperation interlocking system of the target new energy power station in real time, and perform fusion processing on the multi-source status information. Based on the multi-source state information after fusion processing, an electrical topology model of the new energy power station is constructed and dynamically updated. The electrical topology model is used to characterize the real-time electrical connection relationship between all primary equipment, secondary equipment and temporary grounding wires in the new energy power station. Using a knowledge graph reasoning model, based on the electrical topology model and the equipment status data in the multi-source status information, the preset anti-misoperation interlocking rule base is traversed and conflict detected to diagnose logical vulnerabilities in the anti-misoperation interlocking rule base. By combining historical operation sequence data, potential operational risk patterns and rule coverage blind spots are identified through machine learning models to improve the diagnostic results of the aforementioned logical vulnerabilities. By analyzing the time series data of lock status and sensor readings in the multi-source status information, and using a preset predictive diagnostic model, the performance degradation trend and potential failure risk of the physical locks in the anti-misoperation locking system, as well as the performance degradation trend and potential failure risk of the position sensors, are identified. Based on the diagnostic results of the aforementioned logical vulnerabilities, as well as the identification results of the performance degradation trends and potential failure risks, a structured diagnostic report is automatically generated, which includes specific vulnerability locations, risk level assessments, and remediation recommendations.

[0006] Furthermore, the construction of the electrical topology model of the new energy power station based on the fused multi-source state information includes: Extract the unique logical node identifier of each electrical device in the new energy power station, the connection relationship data between devices, and the physical location coordinates of each device from the multi-source state information after fusion processing; Based on the physical location coordinates, each electrical device is mapped to a preset station spatial coordinate system. The preset station spatial coordinate system is a two-dimensional or three-dimensional coordinate system used to characterize the geographical layout of the new energy station and the relative physical location of the equipment. Based on the connection relationship data and the mapping results, an electrical topology model is generated that includes logical node identifiers, device connection relationships, and mapping relationships between physical location coordinates.

[0007] Furthermore, the dynamic updating of the electrical topology model of the new energy power station includes: Real-time monitoring of state change signals related to the topology in the multi-source state information, including switch position signals, protection action signals, grounding wire status signals, and equipment activation / deactivation signals; Count the frequency of occurrence of the state change signal within a specified time window; The update interval of the electrical topology model is adaptively adjusted according to the occurrence frequency; wherein, the higher the occurrence frequency, the shorter the update interval. When the adjusted update interval is triggered, the node status, connection relationship and physical location coordinate mapping in the electrical topology model are synchronously updated according to the state change signal.

[0008] Furthermore, the step of utilizing a knowledge graph reasoning model to traverse and detect conflicts in a pre-set anti-misoperation interlocking rule base based on the electrical topology model and the equipment state data in the multi-source state information includes: The real-time electrical connection relationships represented in the electrical topology model and the switch opening and closing status, protection pressure plate activation and deactivation status, and grounding wire connection and disconnection status in the equipment status data are mapped to entities, attributes, and relationships in the knowledge graph to generate the current system status graph. Each anti-misoperation locking logic in the anti-misoperation locking rule base is parsed into a rule pattern in the knowledge graph. The rule pattern consists of a set of conditional entities, state attribute constraints, and target actions. The rule pattern is traversed and matched in the current system state graph. If a match is successful, the corresponding locking permission judgment is triggered. If contradictory condition entity sets or state attribute constraints corresponding to the same target action are found during the matching process, it is detected as a rule conflict.

[0009] Furthermore, the step of parsing each anti-misoperation locking logic in the anti-misoperation locking rule base into a rule pattern in the knowledge graph includes: Perform syntax parsing on each anti-misoperation interlocking logic and extract the logical expression; The operation objects in the logical expression are converted into entity nodes, the operation states are converted into attribute edges, and the logical relationships are converted into relationship edges to construct an initial rule graph fragment. Based on the device hierarchy and connection specifications defined in the electrical topology model, the initial rule graph fragment is standardized and expanded to supplement the implicit entity relationships and state constraints, forming a complete rule pattern.

[0010] Furthermore, the step of traversing and matching the rule pattern in the current system state graph includes: For the rule pattern that fails to match, determine whether it does not meet the normal locking condition of the current state, or whether it is a potential rule coverage blind spot caused by incomplete rule condition definition. For matching failures identified as rule coverage blind spots, analyze the entities, attributes, or relationships required by the rule pattern but missing in the current system state graph; Based on the missing entities, attributes, or relationships, the matching failure is diagnosed and categorized as a logical vulnerability in the anti-misoperation locking rule base.

[0011] Furthermore, the method of combining historical operational sequence data and using machine learning models to identify potential operational risk patterns and rule coverage blind spots includes: The historical operation sequence data is preprocessed to extract a sequence feature vector containing the operating device, the state before and after the operation, and the operation order; The sequence feature vector is input into a preset operation mode analysis model, which is trained and generated based on normal historical operation data and is used to calculate the deviation between the current operation sequence and the normal operation mode. Operation sequence segments with deviations exceeding a preset threshold are identified as potential operational risk patterns.

[0012] Furthermore, the rule-based blind spot coverage through machine learning models includes: The operation sequences that were successfully executed and did not trigger alarms or interlocks in the anti-misoperation interlocking system were selected from the historical operation sequence data and used as random coverage samples. Analyze the recurring specific device state combinations and operation sequences in the irregularly covered samples; The specific equipment status combinations and operation sequences that conform to the current station operation mode and occur more frequently than a preset threshold are diagnosed as new rule coverage blind spots.

[0013] Furthermore, the method of combining historical operational sequence data and using machine learning models to identify potential operational risk patterns and rule coverage blind spots includes: From the multi-source state information, extract first time-series data characterizing the mechanical action characteristics of the physical lock and second time-series data characterizing the detection accuracy of the position sensor; The first time series data and the second time series data are respectively input into the corresponding predictive diagnostic models; the predictive diagnostic models are generated based on historical health data and fault data, and are used to predict the evolution trend of equipment performance indicators and assess the probability of failure. Based on the output of the predictive diagnostic model, the performance degradation trend and potential failure risk level of the physical lock, as well as the performance degradation trend and potential failure risk level of the position sensor, are determined.

[0014] Furthermore, by combining the diagnostic results of the logical vulnerabilities with the identification results of the performance degradation trends and potential failure risks, a structured diagnostic report is automatically generated, including specific vulnerability locations, risk level assessments, and remediation recommendations, comprising: The diagnosed logic vulnerabilities and identified hardware performance degradation trends and potential failure risks are aggregated to form a set of diagnostic items to be evaluated; Based on preset risk assessment rules, a comprehensive risk score is given to each diagnostic item in the set of diagnostic items to be assessed and the risk level is divided. The risk assessment rules define the electrical impact range and historical trigger frequency of the logic vulnerability, as well as the failure probability of hardware risks and the contribution weight of equipment criticality to the risk score. Based on the risk level and the type of diagnostic item, a pre-set remediation measure knowledge base is matched to generate targeted remediation measure suggestions; The report integrates entries containing specific vulnerability locations, risk device identification, risk level, and remediation recommendations to generate a structured diagnostic report.

[0015] As can be seen from the above technical solutions, the present invention has the following advantages: This invention constructs and dynamically updates a topology model representing the real-time electrical connection relationships of a power station by real-time access and fusion of multi-source state information from power grid data acquisition systems, relay protection devices, smart grounding wires, and environmental sensors. Based on this, on the one hand, a knowledge graph reasoning model is used to traverse and detect conflicts in a static anti-misoperation rule base to diagnose logical vulnerabilities; on the other hand, historical operational data is combined with a machine learning model to identify potential risk patterns and rule blind spots to improve diagnostic results. Predictive models analyze the time-series data of locks and sensors to identify hardware performance degradation trends and fault risks. Finally, by integrating all the above diagnostic and identification results, a structured report containing precise location, risk rating, and repair suggestions is automatically generated. This invention solves the problem that existing systems cannot respond in real time to dynamic risks such as topology changes and backfeeding at renewable energy power plants, improving the situational adaptability and accuracy of error prevention judgments. Secondly, it can proactively discover logical conflicts, redundancies, and coverage blind spots that are difficult for humans to detect, and identify hidden operational risk patterns, significantly improving the reliability of the rule base. Finally, the generated operable structured report provides precise and efficient decision support for operation and maintenance personnel, realizing intelligent closed-loop management for error prevention. It effectively improves the inherent safety level of renewable energy power plants in complex operating environments and the overall adaptive reliability of the system. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating an intelligent diagnostic method for an anti-misoperation interlocking system in a new energy power station according to the present invention. Figure 2 This is a schematic diagram of the electrical topology model for constructing a new energy power station in this invention; Figure 3 This is a flowchart illustrating the process of dynamically updating the electrical topology model of a new energy power station in this invention. Figure 4 This is a flowchart illustrating the process of diagnosing logical vulnerabilities in the anti-misoperation locking rule base in this invention. Figure 5 This is a schematic diagram of the process for identifying potential operational risk patterns and rule coverage blind spots in this invention; Figure 6 This is a schematic diagram of the process of identifying rule coverage blind spots using a machine learning model in this invention; Figure 7 This is a schematic diagram of the process for identifying the performance degradation trend and potential failure risks of physical locks and position sensors in the anti-misoperation interlocking system of the present invention. Figure 8 This is a schematic diagram of the process for generating a structured diagnostic report in this invention. Detailed Implementation

[0017] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0018] Example 1 The implementation method in this embodiment can be implemented in a system, on a server, or on a terminal; no specific limitation is made. The method in this application will be described below from the perspective of system implementation. Please refer to... Figure 1 The method provided in this application includes the following steps: S1. Real-time access to multi-source status information from the power grid data acquisition system, relay protection device, intelligent grounding wire management unit, and environmental monitoring sensor associated with the anti-misoperation interlocking system of the target new energy power station, and fusion processing of the multi-source status information; The multi-source status information here includes telemetry and communication data from the power grid data acquisition system, the protection plate activation / deactivation status and action signals from relay protection devices, the grounding wire connection / deactivation status and unique identification information from the intelligent grounding wire management unit, and environmental parameters such as temperature, humidity, and wind speed from environmental monitoring sensors. All of the above information is continuously collected and transmitted to the data processing unit of the anti-misoperation interlocking system through corresponding communication interfaces and protocols. The system performs fusion processing on heterogeneous status information from different sources with different formats and update frequencies. This fusion processing mainly involves timestamping the data to solve timing synchronization problems, cleaning and verifying the data to remove outliers, and mapping all status information to a common equipment object model and data space, thereby forming a complete panoramic status dataset for the station, providing an accurate and reliable data foundation for all subsequent diagnostic steps.

[0019] S2. Based on the multi-source state information after fusion processing, construct and dynamically update the electrical topology model of the new energy power station. The electrical topology model is used to characterize the real-time electrical connection relationship between all primary equipment, secondary equipment and temporary grounding wires in the new energy power station. The electrical topology model is a dynamic data structure used to describe the real-time connection relationships between all electrical equipment in the site, including primary equipment such as circuit breakers and disconnect switches, secondary equipment such as protection devices, and temporary grounding wires.

[0020] Please see Figure 2 The electrical topology model of the new energy power station is constructed, including the following steps: S211. Extract the unique logical node identifier of each electrical device in the new energy power station, the connection relationship data between devices, and the physical location coordinates of each device from the multi-source state information after fusion processing; S212. Based on the physical location coordinates, map each electrical device to a preset station spatial coordinate system. The preset station spatial coordinate system is a two-dimensional or three-dimensional coordinate system used to characterize the geographical layout of new energy stations and the relative physical location of equipment. S213. Based on the connection relationship data and mapping results, generate an electrical topology model that includes logical node identifiers, device connection relationships and mapping relationships between physical location coordinates.

[0021] Specifically, the unique logical node identifier is a unique code for each device within the system, used to identify the device; the connection relationship data between devices describes the electrical connection paths formed between devices through cables, busbars, etc.; the physical location coordinates are the actual geographical coordinates of the device within the site, which can be obtained from design drawings or positioning equipment. This site spatial coordinate system is established with a fixed reference point within the site as the origin, used to uniformly quantify the device location. Through coordinate transformation, the physical coordinates of each device are mapped to this system, achieving precise spatial positioning of the device. The system integrates the logical node identifier, connection relationship, and physical location coordinates to form a comprehensive data model containing device attributes, connection topology, and spatial information, namely, the electrical topology model.

[0022] Please see Figure 3 Dynamically updating the electrical topology model of new energy power plants includes the following steps: S221. Real-time monitoring of state change signals related to the topology in multi-source status information, including switch position signals, protection action signals, grounding wire status signals, and equipment activation / deactivation signals; S222. Count the frequency of occurrence of state change signals within a specified time window; S223. The update interval of the electrical topology model is adaptively adjusted according to the occurrence frequency; wherein, the higher the occurrence frequency, the shorter the update interval; S224. When the adjusted update interval is triggered, the node status, connection relationship and physical location coordinate mapping in the electrical topology model are updated synchronously according to the state change signal.

[0023] Specifically, state change signals reflect events that may cause changes in the electrical topology. For example, switch position change signals indicate a change in switch state, protection action signals indicate the activation of protection devices, grounding wire status signals indicate changes in grounding wire connection / removal status, and equipment activation / deactivation signals indicate the activation or deactivation of equipment. A specified time window is a preset time period, such as the past five minutes. Within this window, the system accumulates the number of times all state change signals occur, obtaining the occurrence frequency. The update interval is the time period for recalculating and refreshing the model. The system dynamically adjusts the interval based on the occurrence frequency; a shorter interval for faster response to changes when the frequency is high, and a longer interval for reduced computational load when the frequency is low. When an update interval is triggered, the system modifies the node state of the corresponding equipment in the model based on the monitored signals, adjusts the connection relationships between equipment, or updates the physical location coordinate mapping to ensure that the model is consistent with the actual electrical connection state on site in real time.

[0024] S3. Using a knowledge graph reasoning model, based on the electrical topology model and the equipment status data in the multi-source state information, the pre-set anti-misoperation interlocking rule base is traversed and conflict detected to diagnose logical vulnerabilities in the anti-misoperation interlocking rule base. Knowledge graph reasoning models are artificial intelligence models that use graph-structured data for logical representation and reasoning. By diagnosing error-prevention locking rule bases, this model can automatically identify issues such as conflicts, redundancy, and incomplete coverage among rules, thereby improving the rigor and reliability of the rule base. Please refer to [link / reference]. Figure 4 Including the following: S311. Map the real-time electrical connection relationships represented in the electrical topology model and the switch opening and closing status, protection pressure plate activation and deactivation status, and grounding wire connection and disconnection status in the equipment status data to entities, attributes, and relationships in the knowledge graph to generate the current system status graph; In practical implementation, the system abstracts each device in the electrical topology model as an entity node in a knowledge graph, with each entity node having a unique identifier. The electrical connections between devices are mapped to directed or undirected edges between entity nodes. Dynamic information such as the opening and closing status of switches, the activation and deactivation status of protection plates, and the attachment and removal status of grounding wires are mapped to attribute key-value pairs for the corresponding entity nodes. Through this mapping process, a current system state graph is generated, dynamically and intuitively representing the real-time operating status of the entire station in a graph structure.

[0025] S312. Parse each anti-misoperation locking logic in the anti-misoperation locking rule base into a rule pattern in the knowledge graph. The rule pattern consists of a set of conditional entities, state attribute constraints, and target actions. 1. Perform syntax parsing on each anti-misoperation interlocking logic and extract the logical expression; 2. Convert the operation objects in the logical expression into entity nodes, the operation states into attribute edges, and the logical relations into relation edges to construct an initial rule graph fragment; 3. Based on the device hierarchy and connection specifications defined in the electrical topology model, the initial rule graph fragments are standardized and expanded to supplement the implicit entity relationships and state constraints, forming a complete rule pattern.

[0026] Specifically, each anti-misoperation interlocking logic is parsed to extract its logical expression. Anti-misoperation interlocking logic is typically written as a text rule in a domain-specific language or Boolean logic. The system uses a parser to identify the operands, operation states, logical operators, and action instructions in the rule text and converts them into structured logical expressions. For example, the rule that the grounding switch can only be closed when the circuit breaker is open and the line side is de-energized is converted into an expression like (Circuit breaker state = open AND Line side voltage = de-energized) -> (Grounding switch allowed = closed). Operands in the logical expressions are converted into entity nodes, operation states into attribute edges, and logical relationships into relational edges to construct an initial rule graph fragment. The system maps specific devices involved in the logical expressions to entity nodes. The state conditions that devices must satisfy are mapped to attribute constraints that entity nodes must satisfy, forming attribute edges. The logical relationships between devices or the causal relationships implied in the expressions are mapped to relational edges between entity nodes. The local graph structure composed of these elements is the initial rule graph fragment, which abstractly expresses the conditional pattern defined by the rule.

[0027] The initial rule graph fragment may not explicitly specify all implicit conditions. For example, a rule might mention a disconnector switch, but according to electrical topology specifications, it might be connected to a specific bus. Standardization expansion involves supplementing the initial fragment with these implicit entity nodes and relational edges based on information such as device affiliations and electrical connection specifications in the electrical topology model, and performing consistency checks on the state constraints. The standardized graph structure obtained after expansion constitutes the complete rule schema.

[0028] S313. Traverse the matching rule patterns in the current system state graph. If a match is successful, trigger the corresponding locking permission judgment. If during the matching process, if contradictory condition entity sets or state attribute constraints corresponding to the same target action are found, then rule conflict is detected.

[0029] 1. For rule patterns that fail to match, determine whether the normal locking condition of the current state is not met, or whether there is a potential rule coverage blind spot due to incomplete rule condition definition; 2. For matching failures identified as rule coverage blind spots, analyze the entities, attributes, or relationships required by the rule pattern but missing from the current system state graph; 3. Based on missing entities, attributes, or relationships, diagnose and categorize matching failures as logical vulnerabilities in the anti-misoperation locking rule base.

[0030] Specifically, a matching failure refers to the inability to find a subgraph in the current system state graph that completely matches the complete rule pattern defined by a certain rule. A normal locking condition means that the rule itself is correct, but the current site state does not meet its conditions, hence the matching failure; this is the expected safety locking behavior. A potential rule coverage blind spot refers to a combination of potentially risky equipment states and operations that does not exist and is not covered or accurately described by any rule pattern, causing the system to fail to identify the risk; this is a deficiency in the rule base. When the system infers that a matching failure is caused by a coverage blind spot, it analyzes which condition(s) were not met. For example, a rule pattern might require checking the potential of a grounding point, but the potential attribute of that grounding point is not defined or cannot be obtained in the current system state graph; or the rule pattern might implicitly contain an interlocking relationship between two devices, but this relationship does not exist in the current graph's connection relationships. The system will structure and record the analyzed missing information, associate it with the corresponding rule pattern and target action, generate a clear diagnostic record, and identify it as a logical vulnerability of the rule coverage incompleteness type. These types of vulnerabilities indicate that the existing rule base fails to effectively cover certain possible combinations of operating scenarios or device states, necessitating the addition or modification of rules.

[0031] S4. By combining historical operation sequence data, identify potential operational risk patterns and rule coverage blind spots through machine learning models to improve the diagnostic results of logical vulnerabilities; This historical operation sequence data refers to a series of actual switching operation records arranged chronologically by the anti-misoperation interlocking system. Each record includes information such as operation time, operating equipment, pre-operation status, and post-operation status. Analyzing this data using a machine learning model aims to identify operational habits or patterns that, while not violating existing explicit rules, deviate from routine and contain risks, as well as to identify safety scenarios not covered by existing rules. This supplements and improves the logical vulnerabilities discovered earlier based on knowledge graph reasoning, making the diagnostic results more comprehensive.

[0032] Please see Figure 5 By combining historical operational sequence data, machine learning models are used to identify potential operational risk patterns and rule coverage blind spots, including the following steps: S411. Preprocess the historical operation sequence data to extract sequence feature vectors containing the operating equipment, the state before and after the operation, and the operation order; S412. Input the sequence feature vector into the preset operation mode analysis model. The operation mode analysis model is generated based on normal historical operation data and is used to calculate the deviation between the current operation sequence and the normal operation mode. S413. Identify operation sequence segments with deviations exceeding a preset threshold as potential operation risk patterns.

[0033] Specifically, the system first cleans and formats the raw operation logs, abstracting each operation into a data unit containing fields such as the operating device identifier, the device status before the operation, and the device status after the operation. Multiple consecutive operation units are combined into an operation sequence according to the operation time order. Using feature engineering methods, each operation sequence is converted into a set of numerical vectors that can represent its core features, i.e., a sequence feature vector. This vector not only contains specific device and status information but also encodes the sequential relationship of the operations. The preset operation pattern analysis model is trained on a large amount of historical operation sequence data labeled as normal or safe using machine learning algorithms. Through learning, the model can establish an internal representation of standard operation patterns. When a new operation sequence feature vector is input, the model calculates the degree of difference between the feature representation of the sequence and the learned regular patterns. This degree of difference is quantified as a numerical value, i.e., the deviation. A preset threshold is the deviation threshold, which is determined by the model's performance on the validation set or by domain expert experience. When the deviation of an operation sequence exceeds the threshold, it indicates a significant difference between the sequence's operation and the normal safety pattern, potentially implying an unreasonable operation sequence, improper device state transitions, or undefined risky behavior. The system will mark such sequence segments as potential operational risk patterns, providing maintenance personnel with clues for review and intervention.

[0034] Please see Figure 6 The rule coverage blind spots are identified through machine learning models, including the following steps: S421. Select operation sequences that were successfully executed and did not trigger alarms or interlocks in the anti-misoperation interlocking system from the historical operation sequence data, and use them as irregular coverage samples; S422. Analyze the recurring combinations of specific device states and operating sequences in irregularly covered samples; S423. Specific equipment status combinations and operation sequences that conform to the current station operation mode and occur more frequently than a preset threshold are diagnosed as new rule coverage blind spots.

[0035] Specifically, the system traverses historical operation logs, filtering out operation records that were successfully executed without any alarms or interlocking actions from the anti-misoperation interlocking system throughout the entire operation. These operation sequences constitute a set of irregular coverage samples, meaning that the existing anti-misoperation rule base recognizes these operations as safe or fails to identify their potential risks. The system uses a frequent pattern mining algorithm to analyze the set of irregular coverage samples, looking for frequently recurring combinations of device states and fixed operation sequences. For example, it may find that under a certain operating mode, a fixed operation sequence is executed on several devices, but this scenario is not specifically defined in the existing anti-misoperation rules. For the frequently discovered patterns, combined with the current electrical topology model and operating mode of the site, it is determined whether they belong to a potentially recurring operating scenario. If it is determined to be a reasonable scenario, and the frequency of the pattern exceeds a threshold set based on historical data statistical analysis, the system considers this a safety gap that the existing rule base has not effectively covered, and diagnoses it as a new rule coverage blind spot. This suggests that the rule base needs to supplement corresponding anti-misoperation logic for such scenarios.

[0036] S5. Analyze the time series data of lock status and sensor readings in multi-source status information, and identify the performance degradation trend and potential failure risk of physical locks in the anti-misoperation interlocking system, as well as the performance degradation trend and potential failure risk of position sensors, through a preset predictive diagnostic model. This step, by analyzing time-series data reflecting equipment operating characteristics, enables early detection of performance degradation signs and assessment of potential failures, thereby improving the hardware reliability of the entire anti-misoperation interlocking system and preventing the anti-misoperation function from malfunctioning due to hardware failure. Please refer to [link / reference]. Figure 7 It includes the following steps: S511. Extract first time-series data characterizing the mechanical action characteristics of the physical lock and second time-series data characterizing the detection accuracy of the position sensor from the multi-source state information; S512. Input the first time series data and the second time series data into the corresponding predictive diagnostic models respectively; the predictive diagnostic models are generated based on historical health data and fault data, and are used to predict the evolution trend of equipment performance indicators and assess the probability of failure. S513. Based on the output of the predictive diagnostic model, determine the performance degradation trend and potential failure risk level of the physical lock, as well as the performance degradation trend and potential failure risk level of the position sensor.

[0037] Specifically, the first time-series data includes signals that change over time, such as the current curve of the lock drive motor, the action completion time, and the bolt displacement feedback, during each unlocking / locking operation. This data reflects the wear and jamming of the lock's mechanical structure. The second time-series data includes the deviation sequence between repeated detection values ​​and known true positions of the position sensor, signal stability indicators, etc. This data reflects the drift in sensor detection accuracy or the decrease in sensitivity. For the physical lock and position sensor, the system deploys a specially trained predictive diagnostic model. The model uses data from the entire process of equipment gradually deteriorating from a healthy state to failure over a historical period, as well as a large amount of normal data as training samples. The model can learn the pattern of equipment performance indicators deteriorating over time. After inputting the current equipment's time-series data, the model outputs two key results: first, the predicted evolution curve of performance indicators over a future period, i.e., the performance degradation trend; second, the probability estimate of the equipment's functional failure within a specific future time period calculated based on the degree of matching between the current data pattern and known failure patterns. The system determines whether the slope of the performance indicator evolution curve predicted by the model exceeds the allowable range, thereby qualitatively and quantitatively determining the performance degradation trend. Simultaneously, the failure probability values ​​calculated by the model are mapped to a preset risk level system to determine the potential failure risk level. Finally, a health status diagnosis conclusion containing a specific description of the degradation trend and a clear risk level is generated for each monitored physical lock and position sensor.

[0038] S6. Based on the diagnostic results of comprehensive logic vulnerabilities, as well as the identification results of performance degradation trends and potential failure risks, a structured diagnostic report is automatically generated, which includes specific vulnerability locations, risk level assessments, and remediation recommendations.

[0039] This step prioritizes the risks of different types and sources identified in the preceding steps and transforms them into knowledge that can directly guide operational actions. Please refer to [link / reference]. Figure 8 The specific implementation includes the following sub-steps: S611. Aggregate the diagnosed logic vulnerabilities and identified hardware performance degradation trends and potential failure risks to form a set of diagnostic items to be evaluated; S612. Based on the preset risk assessment rules, a comprehensive risk score is given to each diagnostic item in the set of diagnostic items to be assessed, and the risk level is divided. The risk assessment rules define the electrical impact range and historical trigger frequency of the logic vulnerability, as well as the failure probability of hardware risks and the contribution weight of equipment criticality to the risk score. S613. Based on the risk level and the type of diagnostic item, match the pre-set remediation action knowledge base to generate targeted remediation action recommendations; S614. Integrate the entries containing specific vulnerability locations, risk device identification, risk level, and remediation recommendations to generate a structured diagnostic report.

[0040] The system transforms the various logical vulnerabilities diagnosed in steps S3 and S4, as well as the performance degradation trends and fault risk levels determined for each hardware device in step S5, into standardized diagnostic entries. Each entry contains core information such as the problem type, involved objects, and problem description. All entries are aggregated to form a set of diagnostic entries to be evaluated. For logical vulnerability entries, configurable risk assessment rules consider the electrical scope that the vulnerability might affect if triggered and the frequency of similar malfunctions in the past, assigning different weights and calculating risk scores. For hardware risk entries, the rules combine the predicted failure probability of the device and its criticality in the error prevention system to calculate risk scores. Finally, based on the total score of each entry, it is categorized into several preset risk levels. The remediation measures knowledge base is associated with recommended remediation or disposal measures under different problem types and risk levels. The system automatically matches and generates corresponding text suggestions based on the attributes of each diagnostic entry. Finally, the system sorts all the processed diagnostic entries from high to low risk level and organizes them in the form of tables or hierarchical documents to form the final structured diagnostic report. The report lists each identified problem, its location, severity, and the actions that should be taken, providing operations and maintenance managers with comprehensive, intuitive, and actionable decision support.

[0041] It is understood that those skilled in the art can combine various implementation methods in the above embodiments under the guidance of the above examples to obtain technical solutions with multiple implementation methods.

[0042] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent diagnostic method for an anti-misoperation interlocking system in a new energy power station, characterized in that, include: The system can access the multi-source status information of the power grid data acquisition system, relay protection device, intelligent grounding wire management unit and environmental monitoring sensor associated with the anti-misoperation interlocking system of the target new energy power station in real time, and perform fusion processing on the multi-source status information. Based on the multi-source state information after fusion processing, an electrical topology model of the new energy power station is constructed and dynamically updated. The electrical topology model is used to characterize the real-time electrical connection relationship between all primary equipment, secondary equipment and temporary grounding wires in the new energy power station. Using a knowledge graph reasoning model, based on the electrical topology model and the equipment status data in the multi-source status information, the preset anti-misoperation interlocking rule base is traversed and conflict detected to diagnose logical vulnerabilities in the anti-misoperation interlocking rule base. By combining historical operation sequence data, potential operational risk patterns and rule coverage blind spots are identified through machine learning models to improve the diagnostic results of the aforementioned logical vulnerabilities. By analyzing the time series data of lock status and sensor readings in the multi-source status information, and using a preset predictive diagnostic model, the performance degradation trend and potential failure risk of the physical locks in the anti-misoperation locking system, as well as the performance degradation trend and potential failure risk of the position sensors, are identified. Based on the diagnostic results of the aforementioned logical vulnerabilities, as well as the identification results of the performance degradation trends and potential failure risks, a structured diagnostic report is automatically generated, which includes specific vulnerability locations, risk level assessments, and remediation recommendations.

2. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power plants according to claim 1, characterized in that, The construction of the electrical topology model of the new energy power station based on the fused multi-source state information includes: Extract the unique logical node identifier of each electrical device in the new energy power station, the connection relationship data between devices, and the physical location coordinates of each device from the multi-source state information after fusion processing; Based on the physical location coordinates, each electrical device is mapped to a preset station spatial coordinate system. The preset station spatial coordinate system is a two-dimensional or three-dimensional coordinate system used to characterize the geographical layout of the new energy station and the relative physical location of the equipment. Based on the connection relationship data and the mapping results, an electrical topology model is generated that includes logical node identifiers, device connection relationships, and mapping relationships between physical location coordinates.

3. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power stations according to claim 2, characterized in that, The dynamic updating of the electrical topology model of the new energy power station includes: Real-time monitoring of state change signals related to the topology in the multi-source state information, including switch position signals, protection action signals, grounding wire status signals, and equipment activation / deactivation signals; Count the frequency of occurrence of the state change signal within a specified time window; The update interval of the electrical topology model is adaptively adjusted according to the occurrence frequency; wherein, the higher the occurrence frequency, the shorter the update interval. When the adjusted update interval is triggered, the node status, connection relationship and physical location coordinate mapping in the electrical topology model are synchronously updated according to the state change signal.

4. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power plants according to claim 1, characterized in that, The step of utilizing a knowledge graph reasoning model, based on the electrical topology model and equipment state data from the multi-source state information, to traverse and detect conflicts in a pre-set anti-misoperation interlocking rule base includes: The real-time electrical connection relationships represented in the electrical topology model and the switch opening and closing status, protection pressure plate activation and deactivation status, and grounding wire connection and disconnection status in the equipment status data are mapped to entities, attributes, and relationships in the knowledge graph to generate the current system status graph. Each anti-misoperation locking logic in the anti-misoperation locking rule base is parsed into a rule pattern in the knowledge graph. The rule pattern consists of a set of conditional entities, state attribute constraints, and target actions. The rule pattern is traversed and matched in the current system state graph. If a match is successful, the corresponding locking permission judgment is triggered. If contradictory condition entity sets or state attribute constraints corresponding to the same target action are found during the matching process, it is detected as a rule conflict.

5. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power stations according to claim 4, characterized in that, The step of parsing each anti-misoperation locking logic in the anti-misoperation locking rule base into a rule pattern in a knowledge graph includes: Perform syntax parsing on each anti-misoperation interlocking logic and extract the logical expression; The operation objects in the logical expression are converted into entity nodes, the operation states are converted into attribute edges, and the logical relationships are converted into relationship edges to construct an initial rule graph fragment. Based on the device hierarchy and connection specifications defined in the electrical topology model, the initial rule graph fragment is standardized and expanded to supplement the implicit entity relationships and state constraints, forming a complete rule pattern.

6. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power stations according to claim 4, characterized in that, The step of traversing and matching the rule pattern in the current system state graph includes: For the rule pattern that fails to match, determine whether it does not meet the normal locking condition of the current state, or whether it is a potential rule coverage blind spot caused by incomplete rule condition definition. For matching failures identified as rule coverage blind spots, analyze the entities, attributes, or relationships required by the rule pattern but missing in the current system state graph; Based on the missing entities, attributes, or relationships, the matching failure is diagnosed and categorized as a logical vulnerability in the anti-misoperation locking rule base.

7. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power plants according to claim 1, characterized in that, The process of combining historical operational sequence data with machine learning models to identify potential operational risk patterns and rule coverage blind spots includes: The historical operation sequence data is preprocessed to extract a sequence feature vector containing the operating device, the state before and after the operation, and the operation order; The sequence feature vector is input into a preset operation mode analysis model, which is trained and generated based on normal historical operation data and is used to calculate the deviation between the current operation sequence and the normal operation mode. Operation sequence segments with deviations exceeding a preset threshold are identified as potential operational risk patterns.

8. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power stations according to claim 7, characterized in that, The method of identifying rules to cover blind spots using machine learning models includes: The operation sequences that were successfully executed and did not trigger alarms or interlocks in the anti-misoperation interlocking system were selected from the historical operation sequence data and used as random coverage samples. Analyze the recurring specific device state combinations and operation sequences in the irregularly covered samples; The specific equipment status combinations and operation sequences that conform to the current station operation mode and occur more frequently than a preset threshold are diagnosed as new rule coverage blind spots.

9. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power stations according to claim 1, characterized in that, The analysis of time-series data of lock status and sensor readings in the multi-source status information, through a preset predictive diagnostic model, identifies the performance degradation trend and potential failure risk of physical locks in the anti-misoperation locking system, as well as the performance degradation trend and potential failure risk of position sensors, including: From the multi-source state information, extract first time-series data characterizing the mechanical action characteristics of the physical lock and second time-series data characterizing the detection accuracy of the position sensor; The first time series data and the second time series data are respectively input into the corresponding predictive diagnostic models; the predictive diagnostic models are generated based on historical health data and fault data, and are used to predict the evolution trend of equipment performance indicators and assess the probability of failure. Based on the output of the predictive diagnostic model, the performance degradation trend and potential failure risk level of the physical lock, as well as the performance degradation trend and potential failure risk level of the position sensor, are determined.

10. The intelligent diagnostic method for the anti-misoperation interlocking system of new energy power stations according to claim 9, characterized in that, The diagnostic results of the comprehensive analysis of the logical vulnerabilities, along with the identification results of the performance degradation trends and potential failure risks, automatically generate a structured diagnostic report containing specific vulnerability locations, risk level assessments, and remediation recommendations, including: The diagnosed logic vulnerabilities and identified hardware performance degradation trends and potential failure risks are aggregated to form a set of diagnostic items to be evaluated; Based on preset risk assessment rules, a comprehensive risk score is given to each diagnostic item in the set of diagnostic items to be assessed and the risk level is divided. The risk assessment rules define the electrical impact range and historical trigger frequency of the logic vulnerability, as well as the failure probability of hardware risks and the contribution weight of equipment criticality to the risk score. Based on the risk level and the type of diagnostic item, a pre-set remediation measure knowledge base is matched to generate targeted remediation measure suggestions; The report integrates entries containing specific vulnerability locations, risk device identification, risk level, and remediation recommendations to generate a structured diagnostic report.

Images