Full-automatic equipment defect positioning system and method of hydropower station relay protection system

Abstract

The invention discloses a full-automatic equipment defect positioning system and method of a hydropower station relay protection system, and relates to the technical field of power system automation. Comprising a data acquisition module, a data processing module, a defect positioning module, an early warning and alarm module, a database module and a man-machine interaction module. The data acquisition module acquires equipment operation data; the data processing module preprocesses the data, extracts features and preliminarily analyzes a fault range; the defect positioning module is fused with fault tree analysis, an expert system and a neural network to accurately position faults; the early warning and alarm module carries out graded early warning and triggers an emergency plan; and the database module adopts a hybrid architecture to store various data. Defect positioning is realized through the steps of data acquisition, processing, positioning, early warning and interactive display. The method achieves the full-automatic rapid diagnosis, precise positioning and effective processing of the fault of the relay protection system, improves the reliability and stability of the relay protection system, and has remarkable economic and social benefits.

Description

Technical Field

[0001] This invention relates to the field of power system automation technology, specifically to a fully automatic equipment defect location system and method for a hydropower station relay protection system. Background Technology

[0002] In modern power systems, hydropower stations serve as crucial power supply nodes, and their safe and stable operation is of paramount importance to ensuring the reliability of the power grid. Relay protection systems, acting as the "safety guardians" of hydropower station equipment, bear the critical responsibility of rapidly responding to and isolating faulty areas when equipment malfunctions or fails. However, as hydropower stations continue to expand in scale and their equipment complexity increases, the risk of failure within the relay protection systems themselves also rises.

[0003] Currently, the detection and location of defects in relay protection equipment at hydropower stations mainly rely on manual inspections and traditional monitoring technologies. Manual inspections are limited by the professional skills, experience, and working conditions of the inspectors, resulting in low efficiency, high missed detection rates, and untimely fault location, making it difficult to meet the extremely real-time requirements of power system operation and maintenance. Traditional monitoring technologies, on the other hand, are mostly based on single parameter thresholds, lacking comprehensive analysis and in-depth exploration of equipment operating status, failing to effectively identify potential faults, and struggling to quickly and accurately locate the root cause of faults in complex fault scenarios.

[0004] Furthermore, existing technologies suffer from loose connections between various stages of the fault handling process, from fault detection to location and resolution, lacking automated and intelligent linkage mechanisms. This results in long fault handling cycles, easily triggering chain reactions in the power system and causing wider power outages and economic losses. Therefore, developing a system capable of fully automated, rapid, and accurate fault diagnosis and location, and providing effective handling solutions, has become an urgent need to improve the reliability and stability of hydropower station relay protection systems. Summary of the Invention

[0005] The purpose of this invention is to provide a fully automatic equipment defect location system and method for a hydropower station relay protection system. The aim is to achieve fully automatic, rapid and accurate fault cause determination when a fault occurs in the relay protection system, accurately locate the specific faulty equipment, promptly issue relevant information, and provide a scientific and reasonable fault handling method, thereby completely solving the problems of low efficiency, poor accuracy and untimely processing of equipment defect detection and location in the prior art.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: The fully automated equipment defect location system for the relay protection system of a hydropower station includes: The data acquisition module is used to collect operating data of relay protection system equipment and key data of power supply-related equipment; The data processing module is used to preprocess the collected data, extract features, construct fault judgment logic, and conduct preliminary analysis of the fault range. The defect location module includes a fault tree analysis unit, an expert system unit, and a neural network unit, which are used to accurately locate faulty equipment. The early warning and alarm module implements tiered early warnings and alarms based on the severity of the fault, and triggers emergency plans. The database module stores device operation data, historical fault data, device relationship data, and fault handling method data. The human-computer interaction module provides maintenance personnel with a user interface for parameter setting, data query, fault diagnosis results, fault handling methods, and remote control operation.

[0007] The aforementioned data acquisition module employs a high-precision sensor array to collect key data from equipment such as power transformers, circuit breakers, fuses, electrical components, and equipment cabinets. The collected data is transmitted to the data processing module via industrial Ethernet or fiber optic cable. The sensor array has self-diagnosis and adaptive adjustment functions, enabling it to monitor its own working status in real time. When external interference or abnormal performance is detected, it automatically adjusts the acquisition parameters to ensure data accuracy.

[0008] The aforementioned data processing module adopts a distributed computing architecture, utilizing multiple servers to process data in parallel, thereby improving data processing efficiency. Furthermore, during data preprocessing, it employs a deep learning-based outlier detection algorithm to accurately identify and remove abnormal data. When determining component failure, it triggers an automatic inspection process by setting a threshold, filters relevant equipment based on the equipment association table, extracts operational data features for analysis, and progressively determines the operational status of equipment related to the faulty equipment.

[0009] The aforementioned defect localization module's fault tree analysis unit uses the fault phenomenon as the top event and related equipment faults as intermediate and bottom events to construct a refined fault tree model. It locates the fault through top-down logical reasoning and step-by-step troubleshooting. The expert system unit transforms domain expert experience into a rule base. After the system acquires fault data, it gradually narrows down the fault range through rule matching and reasoning. The neural network unit employs transfer learning technology to transfer model parameters trained on fault data from other similar power systems to this system. It then fine-tunes the model by combining a small amount of historical fault data from this hydropower station, quickly constructing a high-precision fault diagnosis model suitable for this system. All three components integrate the diagnostic results through an integrated decision module, improving the accuracy of fault localization.

[0010] The aforementioned early warning and alarm module has a fault trend prediction function. By analyzing the time series of equipment operation data, it can predict the probability and time of fault occurrence in advance, issue early warning information at the fault initiation stage, and set up a graded early warning mechanism according to the severity of the fault to trigger corresponding early warning measures and activate the preset emergency plan.

[0011] The aforementioned database module adopts a hybrid architecture of relational and non-relational databases. The relational database is used to store structured data such as basic device parameter data, historical fault data, and defect case data, while the non-relational database is used to store unstructured data such as real-time monitoring data and raw sensor signals. The database module is equipped with a data encryption storage mechanism, using the AES256 encryption algorithm to encrypt sensitive information. It also has data desensitization tools to desensitize sensitive data when providing data services to external parties, and has automatic data backup and recovery functions.

[0012] The aforementioned human-computer interaction module is developed based on Web technology. The backend uses the Spring Boot framework to build service interfaces, and the frontend uses the Vue.js framework to build a visual interface. It supports multi-language switching and can switch the system interface language in real time according to the settings of the operation and maintenance personnel. When a fault occurs, it displays the location of the faulty device, component code, fault tree analysis process and corresponding fault handling methods, and supports remote operation by operation and maintenance personnel.

[0013] The aforementioned equipment defect location system also has a remote upgrade function, which can remotely download and install software upgrade packages for the data acquisition module, data processing module, defect location module, early warning and alarm module, database module, and human-computer interaction module via the network, so as to achieve continuous optimization and updating of system functions.

[0014] The aforementioned equipment defect location system also includes a security protection module, employing various security technologies such as intrusion detection systems (IDS), firewalls, and security auditing to monitor and protect the system's network communication and data access in real time, preventing unauthorized external intrusion and data leakage.

[0015] The data acquisition module, data processing module, defect location module, early warning and alarm module, database module, and human-computer interaction module mentioned above use a message queue mechanism for data interaction to ensure the reliability and stability of data transmission and avoid data loss and duplicate transmission.

[0016] The data acquisition card in the aforementioned data acquisition module is a high-speed acquisition card with multi-channel synchronous sampling function, a sampling frequency of not less than 10kHz, and an A / D conversion accuracy of 16 bits, ensuring the accuracy and real-time performance of the acquired data.

[0017] The neural network unit of the aforementioned defect localization module uses the TensorFlow framework to build a multi-layer convolutional neural network (CNN) model. The input layer is the extracted equipment operation data features, and the output layer is the fault type and location prediction results. The Adam optimizer and cross-entropy loss function are used to train the model.

[0018] The aforementioned early warning and alarm module uses an industrial-grade audible and visual alarm device, which is connected to the system server via an RS485 communication interface. The alarm is started and stopped by a program. The SMS early warning function is implemented by accessing the Alibaba Cloud Information Service API, and the email early warning uses the JavaMail API to send emails.

[0019] The relational database module described above uses MySQL, configured with an 8-core CPU, 32GB of memory, and 1TB of storage, employing the InnoDB storage engine. The non-relational database uses MongoDB, and a sharded cluster architecture is adopted, deploying 3 shard nodes, 2 configuration server nodes, and 2 arbitration nodes to ensure high availability and scalability of data storage.

[0020] The aforementioned human-computer interaction module uses ECharts to achieve data visualization, including real-time curves of equipment operating parameters, dynamic display of fault trees, and progress charts of fault handling processes. An interactive operation area is designed to allow maintenance personnel to view detailed analysis processes and handling methods by clicking on fault tree nodes, adjust equipment operating parameters by dragging and dropping, and confirm the execution of remote control commands through pop-up windows.

[0021] The defect location method using the fully automatic equipment defect location system of the hydropower station relay protection system described above includes the following steps: S1: The data acquisition module collects real-time operating data of the relay protection system equipment and key data of power-related equipment through a high-precision sensor array and video monitoring equipment. The collected data is converted by A / D and then transmitted to the data processing module via industrial Ethernet or optical fiber. S2: The data processing module preprocesses the collected data, uses algorithms such as median filtering and Gaussian filtering to reduce noise, combines normalization methods to unify the data format and range, and then extracts data features through techniques such as Fourier transform, wavelet transform, and time series analysis to construct fault judgment logic. Based on the equipment association table, it filters relevant equipment and extracts its operating data features, and uses machine learning algorithms to preliminarily analyze the fault range. S3: The fault tree analysis unit, expert system unit, and neural network unit of the defect location module process the analysis results of the data processing module respectively. The fault tree analysis unit constructs a fault tree model for logical reasoning, the expert system unit matches real-time data with the rule base for reasoning, and the neural network unit makes predictions through the trained model. The three units integrate and make decisions to determine the faulty equipment and the cause of the fault. S4: The early warning and alarm module implements graded early warning and alarm based on the diagnostic results of the defect location module and the severity of the fault, triggers the corresponding emergency plan, and updates the fault status information in real time. S5: The human-computer interaction module displays the fault location results, fault causes, and corresponding fault handling methods, supporting maintenance personnel to perform parameter settings, data queries, and remote control operations.

[0022] In step S1 above, the data acquisition module focuses on monitoring equipment related to the power supply of relay protection system components, such as power transformers, circuit breakers, and fuses. 0.2S-class voltage transformers and 0.2-class fiber optic current transformers are installed on the high and low voltage sides of the power transformer. Micro-switch sensors are installed at both ends of the circuit breaker operating mechanism and fuses. High-precision video monitoring cameras are installed in front of the component parts that require visual monitoring.

[0023] In step S2 above, the machine learning algorithms include random forest and support vector machine. When the data processing module judges the component failure, it pre-sets the normal working threshold range based on the component design parameters and actual operating experience. When the monitored data exceeds the threshold, the fully automatic fault diagnosis process is immediately triggered.

[0024] In step S3 above, the integrated decision-making adopts a weighted voting algorithm, which combines the fault tree analysis results, expert system reasoning conclusions and neural network prediction results to obtain the final fault location result.

[0025] In step S4 above, the graded early warning mechanism includes three levels of early warning. When the monitored data drops to 80%-90% of the threshold, a level one early warning is triggered, and a prompt message and email are sent. When the monitored data drops to 70%-80% of the threshold, a level two early warning is triggered, and an audible and visual alarm is added. When the monitored data is below 70% of the threshold, a level three emergency alarm is triggered, all emergency plans are activated, and alarm information is repeatedly pushed through SMS and email.

[0026] This invention discloses a fully automated equipment defect location system and method for a hydropower station relay protection system, which solves the problems of low efficiency and poor accuracy associated with traditional hydropower station relay protection systems that rely on manual inspections and single-parameter threshold monitoring. In comparison, the fully automated equipment defect location system and method of this invention have significant advantages: 1. High efficiency and automation: It realizes full automation of data collection, analysis, location and early warning, quickly responds to fault signals, locates faulty equipment, shortens the fault handling cycle and reduces the risk of chain accidents.

[0027] 2. Precise diagnosis: Combining multi-source data fusion with advanced algorithms, and employing a location method that integrates fault tree analysis, expert systems, and neural networks, the fault range can be accurately located to specific equipment or components, significantly reducing false positives and false negatives, and ensuring stable system operation.

[0028] 3. Intelligent Operation and Maintenance: It has predictive maintenance capabilities, which can provide early warnings of potential faults by analyzing the time series of equipment operation data, helping operation and maintenance personnel to proactively plan maintenance and reduce equipment failure rate and operation and maintenance costs.

[0029] 4. Safe and reliable: The system features a comprehensive security protection design, combining multiple security technologies and redundant backup mechanisms to ensure data security. It also employs a message queue mechanism to ensure reliable data transmission between modules, so that the system can still operate stably even if some modules fail.

[0030] 5. Convenient Interaction: A user-friendly and intelligent human-machine interface supports remote operation and multi-language switching. In case of a fault, the solution is displayed intuitively, which significantly improves the efficiency of operation and maintenance management and reduces the burden on personnel.

[0031] 6. High scalability: The system has remote upgrade capabilities and can continuously optimize and update its functions according to actual needs, adapting to the development and changes of hydropower station relay protection systems. Attached Figure Description

[0032] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a data flow diagram of the present invention; Figure 3 This is a connection diagram of the device of the present invention; Figure 4 This is a flowchart illustrating the logic judgment for component power failure faults according to the present invention. Figure 5 This is a schematic diagram of the fault tree analysis method of the present invention. Detailed Implementation

[0033] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0034] A fully automated equipment defect location system and method for hydropower station relay protection systems, including the following: (I) System Overall Architecture The fully automated equipment defect location system of the hydropower station relay protection system of this invention consists of six core modules: a data acquisition module, a data processing module, a defect location module, an early warning and alarm module, a database module, and a human-machine interaction module. Each module has a clear division of labor yet collaborates closely, forming a complete automated closed-loop system through efficient data interaction and rigorous logical connections. From real-time acquisition of equipment operation data to in-depth data processing and analysis, precise fault location, and timely early warning feedback and processing method output, each link is seamlessly connected, ensuring that the system can respond quickly and operate efficiently when a fault occurs.

[0035] like Figure 1 As shown, the system includes a data acquisition module, a data processing module, a defect location module, an early warning and alarm module, a human-computer interaction module, and a database module. The arrows indicate the data transmission direction. The data acquisition module transmits acquired data to the data processing module, the data processing module transmits processing and analysis results to the defect location module, the defect location module transmits location results to the early warning and alarm module and the human-computer interaction module, and the database module has an interactive relationship with the data processing module, the defect location module, and the human-computer interaction module to retrieve / store data.

[0036] (II) Detailed description of each module 1. Data Acquisition Module The data acquisition module acts as the system's "sensory tentacles," acquiring comprehensive, real-time operational data from the hydropower station's relay protection system equipment. Equipped with a high-precision, high-reliability sensor array, it encompasses various types of sensors, including current transformers, voltage transformers, temperature sensors, humidity sensors, and on-site video signals. Based on equipment operating characteristics and high-fault locations, these sensors are scientifically deployed at key nodes of the relay protection system, ensuring accurate acquisition of core operating parameters such as current, voltage, power, temperature, humidity, switch status, and pressure. Redundant fiber optic communication is employed.

[0037] like Figure 2 As shown, the data acquisition module collects equipment operating data such as current and voltage, and transmits it to the data processing module via the data transmission network. The data processing module performs data preprocessing (noise reduction, normalization), feature extraction (Fourier transform, wavelet transform, etc.), constructs the logic for judging the power failure of components, analyzes relevant equipment data, and preliminarily judges the scope of the fault. The database module provides data support for each module. The defect location module determines the faulty equipment and cause through fault tree analysis, expert system reasoning, and neural network prediction. The early warning and alarm module performs graded early warning / alarm and activates the emergency plan. The human-machine interaction module displays fault information and handling methods to maintenance personnel.

[0038] For equipment related to the power supply of relay protection system components, such as power transformers, circuit breakers, and fuses, key monitoring is implemented. High-resolution and fast-response voltage and current transformers are installed on the power transformers to track the dynamic changes in input and output voltage and current in real time. High-precision auxiliary contact sensors are added to circuit breakers and fuses to capture their opening and closing status signals in real time. The acquired analog signals are converted to digital values ​​by a high-speed, high-precision data acquisition card and then transmitted quickly and stably to the data processing module via industrial Ethernet or fiber optic data transmission networks with millisecond-level transmission cycles and extremely low packet loss rates, providing a reliable data foundation for subsequent fault analysis.

[0039] like Figure 3 As shown, the relay protection equipment group includes current transformers, voltage transformers, temperature sensors, humidity sensors, power transformers, circuit breakers, fuses, etc. These devices are physically connected to the modules of the fully automatic equipment defect location system. The data acquisition module acquires the operating data of each device, and other modules cooperate according to their functions.

[0040] The sensors and video surveillance equipment in the data acquisition module have self-diagnostic and adaptive adjustment functions, enabling them to monitor their own working status in real time. When external interference or abnormal performance is detected, they automatically adjust the acquisition parameters to ensure data accuracy. The data acquisition card is a high-speed card with multi-channel synchronous sampling function, a sampling frequency of no less than 10kHz, and an A / D conversion accuracy of 16 bits, ensuring the accuracy and real-time performance of the acquired data.

[0041] 2. Data Processing Module The data processing module, acting as the system's "intelligent hub," undertakes the in-depth processing and analysis of massive amounts of raw data. It employs a distributed computing architecture, utilizing multiple servers to process data in parallel, thereby improving processing efficiency. First, it uses algorithms such as median filtering and Gaussian filtering to reduce noise in the data. Then, it combines this with normalization methods to unify the data format and range. Simultaneously, it employs a deep learning-based outlier detection algorithm to accurately identify and remove abnormal data, significantly improving data quality.

[0042] like Figure 4 As shown, in the feature extraction stage, machine learning and advanced signal processing algorithms are comprehensively used to deeply mine data features. For current and voltage data, techniques such as Fourier transform and wavelet transform are used to extract key features such as amplitude, phase, and harmonics; for temperature and humidity data, time series analysis and other methods are used to extract features such as changing trends and fluctuation patterns.

[0043] When a fault occurs in the relay protection system, the data processing module quickly initiates the fault diagnosis mechanism. Taking a component power loss fault as an example, a precise normal operating voltage threshold range is pre-set based on the component's design parameters and actual operating experience. Once the component's voltage data is detected to exceed the threshold, a fully automated fault diagnosis process is immediately triggered. Relying on the detailed equipment relationship table in the database module, all relevant equipment that may affect the power supply to the component is automatically screened out. The operating data characteristics of these devices are quickly extracted, such as the output voltage stability and waveform distortion of the power transformer, the opening and closing sequence and state switching frequency of the circuit breaker, and the voltage drop across the fuse. Then, machine learning algorithms such as random forest and support vector machine are used for deep analysis and modeling to preliminarily determine the fault type and possible fault range in a very short time.

[0044] 3. Defect Location Module The defect location module, based on the analysis results of the data processing module, uses a component relationship tree established within the system to perform layer-by-layer analysis, comprehensively judging the status of upstream and downstream equipment of the faulty component, and using a fault logic judgment system to achieve accurate fault location. The defect location module includes a fault tree analysis unit, an expert system unit, and a neural network unit.

[0045] The fault tree analysis unit uses the fault phenomenon as the top event and related equipment faults selected by the data processing module as intermediate and bottom events to construct a refined fault tree model. Through top-down logical reasoning and step-by-step investigation, it deeply analyzes every possible factor that may cause the fault and accurately locates the specific faulty device by combining the determined state of the components themselves.

[0046] The expert system unit leverages the extensive experience and knowledge of domain experts, along with in-depth analysis of numerous historical fault cases, to establish a comprehensive defect diagnosis rule base. When the system detects a fault, it quickly and accurately matches and logically infers real-time collected equipment operating data against rules in the rule base, gradually narrowing down the fault scope until the faulty equipment is located. For example, if the rule base contains the rule "If the power transformer output voltage is zero and the circuit breaker is in the closed state, it may be a transformer fault or a fault in its upstream line," the system automatically retrieves upstream line voltage monitoring data to determine if the upstream line voltage is normal. If the upstream line voltage is normal, the fault scope is narrowed down to the power transformer itself, further analyzing data such as transformer winding current, oil temperature, and oil chromatography to determine if there are faults such as short circuits or overheating inside the transformer. If the upstream line voltage is also zero, the system continues to trace upstream, checking the operating status of other equipment connected to the line, such as upstream circuit breakers and busbars. By sequentially checking the status data of each device, the fault scope is gradually narrowed down.

[0047] The neural network unit constructs a high-precision fault diagnosis model by performing unsupervised or semi-supervised learning on massive amounts of equipment operation data and defect cases containing component failure information. This model can intelligently predict and analyze newly collected equipment operation data, quickly identifying potential fault modes. The neural network unit uses the TensorFlow framework to build a multi-layer convolutional neural network (CNN) model. The input layer is the extracted equipment operation data features, and the output layer is the predicted fault type and location. The model is trained using the Adam optimizer and cross-entropy loss function. Simultaneously, transfer learning technology is employed to transfer model parameters trained on fault data from other similar power systems to this system. Fine-tuning is then performed using a small amount of historical fault data from this hydropower station, rapidly constructing a high-precision fault diagnosis model suitable for this system.

[0048] In practical applications, the fault tree analysis unit, expert system unit, and neural network unit are organically integrated. An integrated decision module uses a weighted voting algorithm to synthesize the diagnostic results of each method, further improving the accuracy and reliability of defect localization and ensuring precise identification of the specific device malfunction that caused the defect. For example... Figure 5 As shown, the top event is component power failure, which is divided into branches such as power supply problems, circuit breaker malfunctions, component problems, environmental factors and other influencing factors. Each branch is further subdivided into specific fault causes. For example, power supply problems include power transformer failures, line failures, fuse blowouts, etc. Power transformer failures are further divided into internal winding short circuits, core failures, tap changer failures, etc. This shows the hierarchical structure of the fault tree and the interrelationships of various fault factors in detail.

[0049] 4. Early Warning and Alarm Module The early warning and alarm module acts as an "information bridge" between the system and maintenance personnel. Based on the diagnostic results of the defect location module, it implements tiered early warning and alarm strategies. The module also features fault trend prediction capabilities, using time-series analysis of equipment operating data to predict the probability and timing of fault occurrences and issue early warning information at the initial stage of a fault.

[0050] Taking component power failure as an example, a three-level early warning mechanism is set up: when the voltage drops to 80%-90% of the threshold, a level one early warning is triggered, and a prompt message and email are sent; when the voltage drops to 70%-80% of the threshold, a level two early warning is triggered, and an audible and visual alarm is added; when the voltage is below 70% of the threshold, a level three emergency alarm is triggered, all emergency plans are activated, and alarm information is repeatedly pushed through SMS and email.

[0051] When components lose power completely, severely impacting normal equipment operation, the system immediately activates the audible and visual alarm device, emitting a strong sound and light signal to attract the attention of on-site personnel. Simultaneously, it sends emergency alarm information to maintenance personnel and automatically triggers pre-set emergency plans, such as activating backup power to maintain critical equipment operation and isolating the fault area to prevent the fault from spreading, minimizing the impact of the fault on the power system. Furthermore, the early warning and alarm module updates fault status information in real time, ensuring that maintenance personnel can promptly understand the progress of fault handling.

[0052] The audible and visual alarm device of the early warning and alarm module is an industrial-grade audible and visual alarm, which is connected to the system server through an RS485 communication interface. The alarm is started and stopped by the program. The SMS early warning function is implemented by accessing the Alibaba Cloud Information Service API, and the email early warning uses the JavaMail API to send emails.

[0053] 5. Database Module The database module, serving as the system's "data repository," employs a hybrid storage architecture combining the relational database MySQL and the non-relational database MongoDB. The MySQL database stores structured data such as basic device parameters, historical fault data, and defect case data. Through a carefully designed table structure and efficient index optimization, it achieves fast data storage and accurate retrieval. The MongoDB database, on the other hand, stores massive amounts of unstructured data generated during device operation, such as real-time monitoring data and raw sensor signals. Leveraging its flexible data model and powerful distributed storage characteristics, it meets the demands for large-scale data storage and high-concurrency access.

[0054] The MySQL database server is configured with an 8-core CPU, 32GB of memory, and 1TB of storage, using the InnoDB storage engine. Tables such as device information, operational data, fault case, and rule base are created, with data relationships constructed through foreign key associations. The MongoDB database employs a sharded cluster architecture, deploying 3 shard nodes, 2 configuration server nodes, and 2 arbitration nodes to ensure high availability and scalability of data storage.

[0055] The database is specifically designed and maintains a detailed equipment relationship table, comprehensively recording information such as power connection relationships, signal transmission relationships, and control logic relationships between various devices in the relay protection system. Simultaneously, a fault handling method database is established, storing handling procedures, operating steps, precautions, and other information for different types of faults, providing data support for fault handling.

[0056] The database module employs an encrypted data storage mechanism, using the AES256 encryption algorithm to encrypt sensitive information such as device operation data and historical fault data. It also includes data masking tools to mask sensitive data before providing data services externally. Furthermore, the database module features automatic data backup and recovery capabilities. Scheduled task scripts perform a full backup of the MySQL database every morning and an incremental backup of the MongoDB database every hour. Backup data is stored on a remote disk array to ensure data integrity, security, and availability.

[0057] 6. Human-Computer Interaction Module The human-computer interaction module provides maintenance personnel with a convenient and intuitive operation and management platform. Developed based on web technologies, the backend uses the Spring Boot framework to build stable and efficient service interfaces, while the frontend uses the Vue.js framework to construct a user-friendly visual interface. Maintenance personnel can easily access the system through a browser to perform functions such as system parameter settings, real-time data queries, historical data statistical analysis, viewing defect diagnosis results, and generating and exporting reports.

[0058] In terms of interface design, ECharts and other chart libraries are fully utilized to display equipment operating data curves, fault tree structures, and defect diagnosis results in intuitive chart and graph formats. ECharts is used to achieve data visualization, including real-time curves of equipment operating parameters, dynamic display of the fault tree, and fault handling process progress charts. An interactive operation area is designed, allowing maintenance personnel to click on fault tree nodes to view detailed analysis processes and handling methods, adjust equipment operating parameters through drag-and-drop operations, and confirm the execution of remote control commands through pop-up windows.

[0059] When a fault occurs, the interface prominently displays the location of the faulty device, component codes, changes in the status of related equipment, and a detailed fault tree analysis and reasoning process. Simultaneously, it retrieves the appropriate handling methods from the fault handling method database and presents them on the interface in a clear, step-by-step, graphical format, allowing maintenance personnel to quickly understand the full picture of the fault and formulate a scientifically sound maintenance plan based on the handling methods. Furthermore, the human-machine interface module supports remote control functionality, enabling maintenance personnel to remotely adjust system parameters and perform remote start-stop, reset, and other operations on equipment, greatly improving the convenience and efficiency of maintenance management.

[0060] The human-computer interaction module supports multi-language switching, and can switch the system interface language in real time according to the settings of the operation and maintenance personnel, including common languages ​​such as Chinese, English, and French, to meet the usage needs of operation and maintenance personnel in different regions.

[0061] (III) Other System Characteristics 1. Remote upgrade function: The system has a remote upgrade function, which can download and install software upgrade packages for the data acquisition module, data processing module, defect location module, early warning and alarm module, database module and human-computer interaction module remotely via the network, so as to realize the continuous optimization and update of system functions.

[0062] 2. Security Protection Function: The system is equipped with a security protection module, which adopts a variety of security technologies such as intrusion detection system (IDS), firewall, and security audit to monitor and protect the system's network communication and data access in real time, preventing unauthorized external intrusion and data leakage.

[0063] 3. Data interaction between modules: The data acquisition module, data processing module, defect location module, early warning and alarm module, database module, and human-computer interaction module use a message queue mechanism for data interaction to ensure the reliability and stability of data transmission and avoid data loss and duplicate transmission.

[0064] (iv) Defect location methods This invention also discloses a fully automatic equipment defect location method for a hydropower station relay protection system, comprising the following steps: S1: The data acquisition module collects real-time operating data of relay protection system equipment and key data of power supply-related equipment through a high-precision sensor array and video monitoring equipment. It focuses on monitoring equipment related to the power supply of relay protection system components, such as power transformers, circuit breakers, and fuses. 0.2S-class voltage transformers and 0.2-class fiber optic current transformers are installed on the high and low voltage sides of the power transformer. Micro-switch sensors are installed at both ends of the circuit breaker operating mechanism and fuses. High-precision video monitoring cameras are installed in front of the components that need to be visually monitored. The collected data is converted by A / D and then transmitted to the data processing module via industrial Ethernet or fiber optic cable. S2: The data processing module preprocesses the collected data, uses median filtering, Gaussian filtering and other algorithms to reduce noise, combines normalization methods to unify the data format and range, uses a deep learning-based outlier detection algorithm to remove abnormal data, and then uses Fourier transform, wavelet transform, time series analysis and other techniques to extract data features, construct fault judgment logic, and pre-set normal operating threshold range based on component design parameters and actual operating experience. When the monitored data exceeds the threshold, the fully automatic fault troubleshooting process is immediately triggered. Based on the equipment association table, relevant equipment is screened and its operating data features are extracted. Machine learning algorithms such as random forest and support vector machine are used to preliminarily analyze the fault range. S3: The fault tree analysis unit, expert system unit, and neural network unit of the defect location module process the analysis results of the data processing module respectively. The fault tree analysis unit constructs a fault tree model for logical reasoning, the expert system unit matches real-time data with the rule base for reasoning, and the neural network unit makes predictions through the trained model. The three units integrate the diagnostic results of each method through a weighted voting algorithm to determine the faulty equipment and the cause of the fault. S4: The early warning and alarm module implements graded early warning and alarm based on the diagnostic results of the defect location module and the severity of the fault. The graded early warning mechanism includes three levels of early warning. When the monitored data drops to 80%-90% of the threshold, a level one early warning is triggered, and a prompt message and email are sent. When the monitored data drops to 70%-80% of the threshold, a level two early warning is triggered, and an audible and visual alarm is added. When the monitored data is below 70% of the threshold, a level three emergency alarm is triggered, all emergency plans are activated, and alarm information is repeatedly pushed through SMS and email. At the same time, the corresponding emergency plan is triggered, and the fault status information is updated in real time. S5: The human-computer interaction module displays the fault location results, fault causes, and corresponding fault handling methods, supporting maintenance personnel to perform parameter settings, data queries, and remote control operations.

[0065] Example 1: (I) Implementation of the data acquisition module High-precision sensors of corresponding types are installed on equipment such as current transformers, voltage transformers, power transformers, circuit breakers, and fuses in the relay protection system of hydropower stations. High-precision video monitoring cameras are installed in front of components requiring visual monitoring. For example, 0.2S-class voltage transformers and 0.2-class fiber optic current transformers are installed on the high and low voltage sides of the power transformer to accurately collect voltage and current data; microswitch sensors are installed at the circuit breaker operating mechanism and both ends of the fuse to monitor their opening and closing status in real time.

[0066] The data acquisition card is a high-speed card with multi-channel synchronous sampling capability, a sampling frequency of no less than 10kHz, and an A / D conversion accuracy of 16 bits, ensuring the accuracy and real-time performance of the acquired data. The acquired digital signals are transmitted to the data processing module via a redundantly designed industrial Ethernet network. The network transmission protocol adopts TCP / IP, and dual network ports are configured for link backup to ensure the stability of data transmission.

[0067] (II) Implementation of the Data Processing Module The data processing server is built with a high-performance multi-core processor, large-capacity memory, and high-speed storage devices, and the operating system is Linux CentOS 7. In the Python 3.8 environment, the Pandas library is used for data reading and preprocessing, including missing value imputation and outlier removal; the StandardScaler function from the Scikit-learn library is used for data normalization.

[0068] In the feature extraction stage, wavelet transform is performed using the PyWavelets library to extract energy features of different frequency bands for current and voltage signals; statistical features such as mean, variance, and kurtosis are calculated using the NumPy library. For the logic implementation of component power failure fault judgment, a Python function is written to set a voltage threshold. When the monitored voltage exceeds the threshold, a database query function based on the SQLAlchemy library is called to retrieve relevant device data from the device association table. Then, a random forest classifier from Scikit-learn is used for preliminary fault diagnosis.

[0069] (III) Implementation of the Defect Location Module Fault tree analysis is implemented by constructing Python classes, defining top, intermediate, and bottom event classes, and building a fault tree model through class inheritance and composition. The Graphviz library is used to visualize the fault tree model, facilitating analysis and debugging. The expert system is developed based on the Drools rule engine, transforming domain expert experience into Drools rule files. After acquiring fault data, the system uses the rule engine for matching and reasoning, gradually narrowing down the fault scope.

[0070] The neural network was built using the TensorFlow framework, constructing a multi-layer convolutional neural network (CNN) model. The input layer consisted of extracted equipment operation data features, and the output layer contained the predicted fault type and location. The model was trained using the Adam optimizer and cross-entropy loss function, with training data derived from historical fault cases and simulated fault data. In practical applications, the fault tree analysis results, expert system inference conclusions, and neural network prediction results were input into the integrated decision module, and the final fault location result was obtained through a weighted voting algorithm.

[0071] (iv) Implementation of the early warning and alarm module The early warning and alarm system is deployed on a dedicated server and developed using Java. SMS alerts are implemented via the Alibaba Cloud Information Service API, while email alerts are sent using the JavaMail API. Industrial-grade audible and visual alarm devices are used, connected to the system server via an RS485 communication interface, and their activation and deactivation are controlled by a Java program.

[0072] To address component power failures, a three-tiered early warning mechanism is implemented: when the voltage drops to 80%-90% of the threshold, a level one early warning is triggered, sending a notification message and email; when the voltage drops to 70%-80% of the threshold, a level two early warning is triggered, adding an audible and visual alarm; and when the voltage falls below 70% of the threshold, a level three emergency alarm is triggered, activating all emergency plans and repeatedly pushing alarm information via SMS and email.

[0073] (v) Database module implementation The MySQL database server is configured with an 8-core CPU, 32GB of memory, and 1TB of storage, using the InnoDB storage engine. Tables such as device information, operational data, fault case, and rule base are created, with data relationships constructed through foreign key associations. The MongoDB database employs a sharded cluster architecture, deploying 3 shard nodes, 2 configuration server nodes, and 2 arbitration nodes to ensure high availability and scalability of data storage.

[0074] Write a scheduled script to perform a full backup of the MySQL database every morning and an incremental backup of the MongoDB database every hour. Backup data is stored on a remote disk array. Data security is ensured through a database firewall, user access control, and data encryption technologies (such as AES encryption).

[0075] (vi) Implementation of the human-computer interaction module The backend is developed using the Spring Boot 2.6 framework to create RESTful API interfaces, employing JWT for user authentication and authorization. The frontend uses Vue 3 + Element Plus to build a single-page application (SPA), interacting with the backend API via the Axios library. ECharts 5 is used for data visualization, including real-time curves of device operating parameters, dynamic display of fault trees, and progress charts of fault handling processes.

[0076] The fault handling interface features an interactive operation area. Maintenance personnel can click on fault tree nodes to view detailed analysis processes and handling methods, adjust equipment operating parameters through drag-and-drop operations, and confirm remote control commands via pop-up windows. The system supports multilingual switching and data export (Excel, PDF formats), enhancing the user experience.

[0077] (vii) System security and upgrade implementation Deploy an intrusion detection system (IDS) and firewall to monitor and protect system network ports and data transmission protocols, preventing unauthorized access and attacks. Establish security audit logs to record system operations, data access, and other behaviors, facilitating the tracing and investigation of security issues.

[0078] Develop a remote upgrade management platform for the system, supporting functions such as upgrade package upload, version verification, and incremental upgrades. Maintenance personnel submit upgrade requests through a human-computer interaction module. The system automatically downloads the upgrade package, verifies its integrity, and upgrades each module in a preset order. Backup data is retained during the upgrade process to ensure a rollback to the original version in case of upgrade failure.

[0079] (viii) Implementation of data interaction between modules RabbitMQ message queues are used as the middleware for data interaction between modules, defining a unified data transmission format and protocol. Each module acts as a message producer or consumer, sending and receiving data through the message queue to achieve asynchronous communication. Message persistence, retry mechanisms, and dead-letter queues are configured to ensure that data is not lost or duplicated during transmission, improving the reliability and stability of the system.

Claims

1. A fully automatic equipment defect location system for a hydropower station relay protection system, characterized in that, include: The data acquisition module is used to collect operating data of relay protection system equipment and key data of power supply-related equipment; The data processing module is used to preprocess the collected data, extract features, construct fault judgment logic, and conduct preliminary analysis of the fault range. The defect location module includes a fault tree analysis unit, an expert system unit, and a neural network unit, which are used to accurately locate faulty equipment. The early warning and alarm module implements tiered early warnings and alarms based on the severity of the fault, and triggers emergency plans. The database module stores device operation data, historical fault data, device relationship data, and fault handling method data. The human-computer interaction module provides maintenance personnel with a user interface for parameter setting, data query, fault diagnosis results, fault handling methods, and remote control operation.

2. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The data acquisition module employs a high-precision sensor array to collect data from power transformers, circuit breakers, fuses, electrical components, and equipment cabinets. The collected data is transmitted to the data processing module via industrial Ethernet or fiber optic cable. The sensor array has self-diagnosis and adaptive adjustment functions, enabling it to monitor its own working status in real time. When external interference or abnormal performance is detected, it automatically adjusts the acquisition parameters to ensure data accuracy.

3. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The data processing module adopts a distributed computing architecture, using multiple servers to process data in parallel, thereby improving data processing efficiency. In the data preprocessing process, a deep learning-based outlier detection algorithm is used to accurately identify and remove abnormal data. When judging component failure, an automatic inspection process is triggered by setting a threshold. Based on the equipment association table, relevant equipment is screened and operational data features are extracted for analysis, and the operational status of equipment related to the faulty equipment is judged layer by layer.

4. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The fault tree analysis unit of the defect localization module uses the fault phenomenon as the top event and related equipment faults as intermediate and bottom events to construct a refined fault tree model. It locates the fault through top-down logical reasoning and step-by-step troubleshooting. The expert system unit transforms the experience of domain experts into a rule base. After the system acquires fault data, it gradually narrows down the fault range through rule matching and reasoning. The neural network unit uses transfer learning technology to transfer the model parameters trained on fault data from other similar power systems to this system. It is then fine-tuned by combining historical fault data from this hydropower station to construct a high-precision fault diagnosis model suitable for this system. The three units integrate the diagnostic results through the integrated decision module to improve the accuracy of fault localization.

5. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The aforementioned early warning and alarm module has a fault trend prediction function. By analyzing the time series of equipment operation data, it can predict the probability and time of fault occurrence in advance, issue early warning information at the fault initiation stage, and set up a graded early warning mechanism according to the severity of the fault to trigger corresponding early warning measures and activate the preset emergency plan.

6. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The database module adopts a hybrid architecture of relational and non-relational databases. The relational database is used to store structured data such as basic device parameter data, historical fault data, and defect case data, while the non-relational database is used to store unstructured data such as real-time monitoring data and raw sensor signals. The database module is equipped with a data encryption storage mechanism, using the AES256 encryption algorithm to encrypt sensitive information. It also has a data desensitization tool to desensitize sensitive data when providing data services to external parties, and has automatic data backup and recovery functions.

7. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The human-computer interaction module is developed based on Web technology. The backend uses the Spring Boot framework to build service interfaces, and the frontend uses the Vue.js framework to build a visual interface. It supports multi-language switching and can switch the system interface language in real time according to the settings of the operation and maintenance personnel. When a fault occurs, it displays the location of the faulty device, component code, fault tree analysis process and corresponding fault handling methods, and supports remote operation by operation and maintenance personnel.

8. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The equipment defect location system also has a remote upgrade function, which allows for the remote download and installation of software upgrade packages for the data acquisition module, data processing module, defect location module, early warning and alarm module, database module, and human-computer interaction module via the network, thereby enabling continuous optimization and updates of system functions.

9. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The equipment defect location system also includes a security protection module, which employs various security technologies such as intrusion detection system (IDS), firewall, and security auditing to monitor and protect the system's network communication and data access in real time, preventing unauthorized external intrusion and data leakage.

10. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The data acquisition module, data processing module, defect location module, early warning and alarm module, database module, and human-computer interaction module use a message queue mechanism for data interaction to ensure the reliability and stability of data transmission and avoid data loss and duplicate transmission.

11. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The data acquisition card in the data acquisition module is a high-speed acquisition card with multi-channel synchronous sampling function, a sampling frequency of not less than 10kHz, and an A / D conversion accuracy of 16 bits, to ensure the accuracy and real-time performance of the acquired data.

12. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The neural network unit of the defect localization module uses the TensorFlow framework to build a multi-layer convolutional neural network (CNN) model. The input layer is the extracted equipment operation data features, and the output layer is the fault type and location prediction results. The model is trained using the Adam optimizer and the cross-entropy loss function.

13. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The aforementioned early warning and alarm module uses an industrial-grade audible and visual alarm device, which is connected to the system server via an RS485 communication interface. The alarm is started and stopped by a program. The SMS early warning function is implemented by accessing the Alibaba Cloud Information Service API, and the email early warning uses the JavaMail API to send emails.

14. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The relational database module uses MySQL, configured with an 8-core CPU, 32GB of memory, and 1TB of storage, employing the InnoDB storage engine. The non-relational database uses MongoDB, and a sharded cluster architecture is adopted, deploying 3 shard nodes, 2 configuration server nodes, and 2 arbitration nodes to ensure high availability and scalability of data storage.

15. The fully automatic equipment defect location system for the hydropower station relay protection system according to claim 1, characterized in that, The human-computer interaction module uses ECharts to visualize data, including real-time curves of equipment operating parameters, dynamic display of fault trees, and progress charts of fault handling processes. An interactive operation area is designed so that maintenance personnel can click on fault tree nodes to view detailed analysis processes and handling methods, adjust equipment operating parameters by dragging and dropping, and confirm the execution of remote control commands through pop-up windows.

16. A defect location method using the fully automatic equipment defect location system of the hydropower station relay protection system according to any one of claims 1-15, characterized in that, Includes the following steps: S1: The data acquisition module collects real-time operating data of the relay protection system equipment and key data of power-related equipment through a high-precision sensor array and video monitoring equipment. The collected data is converted by A / D and then transmitted to the data processing module via industrial Ethernet or optical fiber. S2: The data processing module preprocesses the collected data, uses algorithms such as median filtering and Gaussian filtering to reduce noise, combines normalization methods to unify the data format and range, and then extracts data features through techniques such as Fourier transform, wavelet transform, and time series analysis to construct fault judgment logic. Based on the equipment association table, it filters relevant equipment and extracts its operating data features, and uses machine learning algorithms to preliminarily analyze the fault range. S3: The fault tree analysis unit, expert system unit, and neural network unit of the defect location module process the analysis results of the data processing module respectively. The fault tree analysis unit constructs a fault tree model for logical reasoning, the expert system unit matches real-time data with the rule base for reasoning, and the neural network unit makes predictions through the trained model. The three units integrate and make decisions to determine the faulty equipment and the cause of the fault. S4: The early warning and alarm module implements graded early warning and alarm based on the diagnostic results of the defect location module and the severity of the fault, triggers the corresponding emergency plan, and updates the fault status information in real time. S5: The human-computer interaction module displays the fault location results, fault causes, and corresponding fault handling methods, supporting maintenance personnel to perform parameter settings, data queries, and remote control operations.

17. The fully automatic equipment defect location method for a hydropower station relay protection system according to claim 16, characterized in that, In step S1, the data acquisition module monitors the power supply of equipment related to the power supply of relay protection system components, such as power transformers, circuit breakers, and fuses. 0.2S-class voltage transformers and 0.2-class fiber optic current transformers are installed on the high and low voltage sides of the power transformer. Micro-switch sensors are installed at both ends of the circuit breaker operating mechanism and fuses. High-precision video monitoring cameras are installed in front of the component parts that require visual monitoring.

18. The fully automatic equipment defect location method for a hydropower station relay protection system according to claim 16, characterized in that, In step S2, the machine learning algorithms include random forest and support vector machine. When the data processing module judges the component failure, it pre-sets the normal working threshold range based on the component design parameters and actual operating experience. When the monitored data exceeds the threshold, the fully automatic fault diagnosis process is immediately triggered.

19. The fully automatic equipment defect location method for a hydropower station relay protection system according to claim 16, characterized in that, In step S3, the integrated decision-making adopts a weighted voting algorithm, which combines the fault tree analysis results, expert system reasoning conclusions, and neural network prediction results to obtain the final fault location result.

20. The fully automatic equipment defect location method for a hydropower station relay protection system according to claim 16, characterized in that, In step S4, the graded early warning mechanism includes three levels of early warning. When the monitored data drops to 80%-90% of the threshold, a level one early warning is triggered, and a prompt message and email are sent. When the monitored data drops to 70%-80% of the threshold, a level two early warning is triggered, and an audible and visual alarm is added. When the monitored data is below 70% of the threshold, a level three emergency alarm is triggered, all emergency plans are activated, and alarm information is repeatedly pushed through SMS and email.

Images