A self-diagnosis and rapid fault location method for hydropower station monitoring systems

Abstract

This invention discloses a self-diagnosis and rapid fault location method for a hydropower station monitoring system, comprising the following steps: Step 1: Full-dimensional data acquisition and preprocessing of the monitoring system; Step 2: Multimodal feature fusion and extraction; Step 3: Hierarchical self-diagnosis verification; Step 4: Rapid fault location; Step 5: Fault level classification and emergency response linkage; Step 6: Adaptive optimization of the diagnostic model; Step 7: Diagnosis and fault log retention. Employing a hierarchical diagnostic mode that links the edge and cloud, combined with multimodal feature fusion, it enables comprehensive self-diagnosis of the monitoring system's own health status, effectively reducing false alarms and false negatives, while balancing real-time performance and accuracy. Based on fault causal graphs and Bayesian network correlation analysis algorithms, combined with the system topology, it can quickly locate the specific location of the fault source, the faulty component, and the triggering factors, solving the problems of delayed fault location and large errors in existing technologies.

Description

Technical Field

[0001] This invention belongs to the field of hydropower station monitoring technology, specifically a method for self-diagnosis and rapid fault location in hydropower station monitoring systems. Background Technology

[0002] The monitoring system of a hydropower station is the core support for ensuring the safe, stable, and efficient operation of the station. It is responsible for real-time monitoring, control, and scheduling of the operating status of various equipment, including turbine generators, speed control systems, excitation systems, and water conveyance systems. With the continuous improvement of the automation and intelligence level of hydropower stations, the structure of the monitoring system is becoming increasingly complex, the number of devices is increasing, and data interaction is becoming more frequent, leading to a significant increase in the probability of system failures and equipment malfunctions.

[0003] Existing diagnostic and fault location methods for hydropower station monitoring systems mostly rely on manual inspections and single-dimensional alarm analysis, which have many shortcomings:

[0004] First, its self-diagnostic capabilities are weak, and it can only identify simple faults in a single device, failing to achieve comprehensive self-diagnosis at the system level. Furthermore, its diagnostic accuracy is low, with prominent false alarms and missed alarms.

[0005] Second, fault location is lagging. Traditional methods require manual inspection of alarm signals one by one, which is inefficient and often fails to detect the source of the fault in time, leading to the expansion of the fault and affecting the normal operation of the hydropower station.

[0006] Third, it lacks the ability to fuse multimodal features, making it impossible to effectively combine equipment operation data, status data and environmental data for comprehensive diagnosis, and making it difficult to identify multi-device coupled faults and hidden faults.

[0007] Fourth, the diagnostic model has poor adaptability. The monitoring systems of hydropower stations of different sizes are quite different, and the existing methods are difficult to adapt flexibly. Moreover, the model cannot adaptively optimize according to changes in the system's operating status.

[0008] Fifth, it relies on human experience, and the accuracy of fault diagnosis and location is greatly affected by the professional level of maintenance personnel, making it difficult to achieve standardized and intelligent operation and maintenance.

[0009] Furthermore, some existing fault diagnosis methods employ a single "edge acquisition-cloud diagnosis" model, which, limited by network bandwidth in remote hydropower station areas, suffers from excessively high diagnostic latency, failing to meet the rapid response needs for emergency faults. Simultaneously, some methods lack a comprehensive fault causal correlation system, making it difficult to accurately locate the fault source and analyze its root cause. Therefore, developing a method capable of comprehensive self-diagnosis, rapid and accurate fault location, strong adaptability, and high intelligence has become an urgent technical challenge in the current operation and maintenance of hydropower station monitoring systems.

[0010] Based on this, a self-diagnosis and rapid fault location method for hydropower station monitoring systems is designed. Summary of the Invention

[0011] In view of the above situation and to overcome the shortcomings of the prior art, the present invention provides a self-diagnosis and rapid fault location method for a hydropower station monitoring system, which effectively solves the problems mentioned in the background art.

[0012] To achieve the above objectives, the present invention provides the following technical solution: a self-diagnosis and rapid fault location method for a hydropower station monitoring system, comprising the following steps:

[0013] Step 1: Full-dimensional data collection and preprocessing of the monitoring system. Collect the operation data, status data and environmental data of equipment at all levels of the hydropower station monitoring system. Standardize, denoise and complete the collected data to obtain standardized data.

[0014] Step 2: Multimodal feature fusion and extraction. Based on the preprocessed standardized data, equipment operation features, abnormal status features, and environmental interference features are extracted respectively. Multimodal feature fusion is completed through feature weight allocation algorithm to obtain fused feature set.

[0015] Step 3: Hierarchical self-diagnosis verification. Construct a lightweight self-diagnosis model at the edge and a cloud-based fusion self-diagnosis model. The edge model performs a preliminary self-diagnosis on the fusion feature set, and the cloud model performs a secondary verification on the edge diagnosis results, outputting the self-diagnosis results, which include whether the system is normal, abnormal, and the type of abnormality.

[0016] Step 4: Rapid fault location. When the self-diagnosis result indicates a system anomaly, based on the preset hydropower station monitoring system topology and fault cause-effect graph, combined with the anomaly features of the fused feature set, the specific location of the fault source, the faulty component, and the fault triggering factors are located through the correlation analysis algorithm.

[0017] Step 5: Fault Level Classification and Emergency Response. Based on the impact range of the fault source, the severity of the fault, and the speed of fault development, fault levels are classified, and corresponding emergency response commands are triggered based on different fault levels to achieve rapid fault handling.

[0018] Step Six: Adaptive Optimization of Diagnostic Model. Based on self-diagnosis results, fault location results, and fault handling feedback data, the parameters of the lightweight self-diagnosis model at the edge and the cloud-integrated self-diagnosis model are dynamically adjusted to improve the model's diagnostic accuracy.

[0019] Step 7: Diagnosis and Fault Log Retention. Record the self-diagnosis process, fault location results, fault handling process, and model optimization parameters to form a complete diagnosis and fault log for subsequent traceability and analysis.

[0020] Preferably, in step one, the operating data includes real-time operating parameters of the hydropower station's turbine generator set, speed regulation system, excitation system, water conveyance system, and power distribution system.

[0021] The status data includes the start / stop status of each device, fault alarm status, and component wear status.

[0022] The environmental data includes temperature, humidity, vibration intensity, and electromagnetic interference intensity inside the hydropower station powerhouse.

[0023] The preprocessing specifically includes: standardizing the data using a standardization algorithm, removing noise interference from the data using a denoising algorithm, and completing missing data using a data completion algorithm to ensure the integrity and accuracy of the standardized data.

[0024] Preferably, in step two, the equipment operating characteristics include the fluctuation amplitude, rate of change, and steady-state deviation of the operating parameters;

[0025] The abnormal status characteristics include the triggering frequency, duration, and number of associated signals of the alarm signals;

[0026] The environmental interference characteristics include the deviation of environmental parameters from standard thresholds and the duration of interference.

[0027] The feature weight allocation algorithm adopts the analytic hierarchy process (AHP) to allocate weights based on the degree of influence of each feature on fault diagnosis. After the weight allocation is completed, the fused feature set is obtained by weighted summation.

[0028] Preferably, in step three, the edge-end lightweight self-diagnosis model adopts a pruned convolutional neural network model to achieve local real-time diagnosis;

[0029] The cloud-based fusion self-diagnostic model adopts a fusion model combining Transformer and Support Vector Machine to perform secondary verification of the diagnostic results at the edge.

[0030] The abnormal types of the self-diagnostic results include equipment hardware failure, software program failure, data transmission failure, and environmental interference abnormalities.

[0031] Preferably, in step four, the preset hydropower station monitoring system topology includes the hierarchical relationship between the equipment layer, transmission layer, monitoring layer and application layer, and clarifies the connection method and data interaction path between each device and module;

[0032] The fault cause-effect graph is constructed based on common fault cases and equipment operation mechanisms of hydropower station monitoring systems, and includes the correlation between fault types, fault characteristics, fault causes and fault impact range.

[0033] The correlation analysis algorithm employs a Bayesian network algorithm, which achieves accurate location of the fault source by matching the abnormal features of the fused feature set with the fault causal graph.

[0034] Preferably, in step five, the fault level is divided into four levels: Level 1 fault is a minor abnormality of a single non-critical device, which does not affect the overall operation of the system; Level 2 fault is a minor abnormality of a single critical device or multiple non-critical devices, which has a slight impact on the operation of the system; Level 3 fault is multiple critical devices or a single core device, which has a serious abnormality, affecting the normal operation of some functions of the system; Level 4 fault is a serious abnormality of the core device or a multi-system linkage fault, which may lead to system shutdown.

[0035] The emergency response commands include equipment start / stop commands, parameter adjustment commands, alarm notification commands, and operation and maintenance scheduling commands. Different fault levels correspond to different emergency response commands.

[0036] Preferably, in step six, the dynamic adjustment of model parameters specifically includes: adjusting the feature weight allocation coefficients based on the deviation between the fault location results and the actual fault; optimizing the model's loss function and activation function based on diagnostic false alarms and missed alarms; supplementing training samples and updating the model's fault identification library based on newly emerging fault types to ensure that the model adaptively adapts to changes in the system's operating state.

[0037] Preferably, in step seven, the diagnosis and fault log includes data acquisition time, preprocessing parameters, fusion feature set, self-diagnosis results, fault location information, fault level, emergency response instructions, handling results, and model optimization parameters. The log is stored in an encrypted manner, and the storage period meets the traceability requirements of hydropower station operation and maintenance. It supports retrieval by time, fault type, and equipment type.

[0038] Preferably, the method further includes a fault retrospective analysis step, which, based on the diagnosis and fault log, retrospectively analyzes the faults that have been handled, analyzes the root causes, development patterns and shortcomings in the handling process, and generates a fault retrospective report for optimizing the fault cause-effect graph and self-diagnosis model.

[0039] Preferably, the method is applicable to monitoring systems of hydropower stations of different sizes. By adjusting the data acquisition range, feature dimensions and model parameters, it can be adapted to the monitoring needs of small, medium and large hydropower stations. The adaptation process does not require modification of the monitoring system hardware.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] 1. The self-diagnosis of this invention is comprehensive and accurate. It adopts a hierarchical diagnosis mode that links the edge and the cloud, and combines multimodal feature fusion to achieve comprehensive self-diagnosis of the health status of the monitoring system itself, effectively reducing false alarms and false negatives, and taking into account the real-time performance and accuracy of self-diagnosis.

[0042] 2. The fault location of this invention is fast and accurate. Based on the fault cause-effect graph and Bayesian network correlation analysis algorithm, combined with the system topology, it can quickly locate the specific location of the fault source, the faulty component and the triggering factor, thus solving the problems of lagging fault location and large error in the existing technology.

[0043] 3. This invention has strong adaptability and a high degree of intelligence. By dynamically adjusting the model parameters, it can be adapted to monitoring systems of hydropower stations of different scales without the need to modify the monitoring system hardware. The diagnostic model can adaptively optimize according to changes in the system's operating status without manual intervention, achieving standardized and intelligent operation and maintenance, and reducing reliance on the professional level of operation and maintenance personnel.

[0044] 4. This invention has strong traceability, complete diagnostic and fault log retention, and optional fault backtracking analysis, which can realize full-process fault tracing, facilitate the summarization of fault patterns, optimize operation and maintenance strategies, and further improve the operation and maintenance level of hydropower station monitoring systems. Attached Figure Description

[0045] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0046] In the attached diagram:

[0047] Fig. 1 This is an overall flowchart of the method of the present invention;

[0048] Fig. 2 This is a block diagram of the fault location logic of the present invention;

[0049] Fig. 3 This is the multimodal feature fusion logic diagram of the present invention; Detailed Implementation

[0050] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0051] Depend on Figs. 1-3 This invention relates to a self-diagnosis and rapid fault location method for a hydropower station monitoring system, comprising the following steps:

[0052] Step 1: Comprehensive Data Acquisition and Preprocessing by the Monitoring System

[0053] This system is used to collect complete data from all levels and devices of the hydropower station monitoring system. Preprocessing eliminates data noise and missing data, providing high-quality data support for subsequent self-diagnosis and fault location. The specific operation is as follows:

[0054] 1) Data Acquisition: Data acquisition modules are deployed at the equipment layer, transmission layer, monitoring layer, and application layer of the hydropower station monitoring system. The acquired data types include three categories: first, operational data, covering real-time operating parameters of the turbine generator units, speed control system, excitation system, water conveyance system, and power distribution system; second, status data, including the start-up and shutdown status, fault alarm status, and component wear status of each piece of equipment; and third, environmental data, including temperature, humidity, vibration intensity, and electromagnetic interference intensity within the hydropower station powerhouse. All acquired data is timestamped to ensure data timeliness.

[0055] 2) Data Preprocessing: A three-step processing approach of standardization, denoising, and data completion is adopted to ensure the integrity and accuracy of the data. First, a standardization algorithm is used to standardize all collected data, converting data of different magnitudes and units into standardized data of a unified standard to eliminate the influence of units. Second, a denoising algorithm is used to remove noise interference from the data, filtering out noise data caused by electromagnetic interference, acquisition errors, etc. Finally, a data completion algorithm is used to complete the missing data that occurred during the acquisition process, ensuring the integrity of the standardized data.

[0056] Step 2: Multimodal Feature Fusion and Extraction

[0057] This method is used to extract multimodal information reflecting system operating status and fault characteristics from preprocessed standardized data, and improves the effectiveness of features through feature fusion, thus solving the problem of low diagnostic accuracy of single features. The specific operation is as follows:

[0058] 1) Single-modal feature extraction: Extract features corresponding to three types of data respectively: First, equipment operation features, including the fluctuation amplitude, rate of change and steady-state deviation of operating parameters; second, abnormal status features, including the triggering frequency, duration and number of associated signals of alarm signals; and third, environmental interference features, including the deviation of environmental parameters from standard thresholds and the duration of interference.

[0059] 2) Multimodal feature fusion: The extracted single-modal features are weighted using the analytic hierarchy process (AHP). The weights are set according to the degree of influence of each feature on fault diagnosis. After the weights are assigned, all single-modal features are fused into a fused feature set by weighted summation, providing comprehensive and effective feature support for subsequent self-diagnosis.

[0060] Step 3: Hierarchical self-diagnosis verification

[0061] A hierarchical diagnostic model that links the edge and cloud is adopted, balancing the real-time nature and accuracy of self-diagnosis, thus solving the problem that existing single-diagnosis models cannot simultaneously satisfy "real-time diagnosis" and "precise diagnosis"; the specific operation is as follows:

[0062] 1) Model Construction: Construct a lightweight self-diagnostic model at the edge and a cloud-based fusion self-diagnostic model; the lightweight self-diagnostic model at the edge adopts a pruned convolutional neural network model to adapt to the low computing power requirements of the hydropower station edge acquisition equipment and is used to realize local real-time diagnosis; the cloud-based fusion self-diagnostic model adopts a fusion model combining Transformer and Support Vector Machine to perform secondary verification of the edge diagnosis results and improve the accuracy of self-diagnosis.

[0063] 2) Hierarchical Diagnosis and Verification: First, the lightweight self-diagnosis model at the edge receives the fused feature set, performs a preliminary self-diagnosis of the monitoring system's operating status, and outputs preliminary diagnostic results. Second, the edge uploads the preliminary diagnostic results and the fused feature set to the cloud. The cloud-based fused self-diagnosis model performs a secondary verification of the preliminary diagnostic results, correcting false alarms and missed alarms in the edge diagnosis by comparing the fused feature set with a preset fault feature library. Finally, the cloud outputs the final self-diagnosis results, which include system normal, system abnormal, and abnormality type. The abnormality type covers four categories: equipment hardware failure, software program failure, data transmission failure, and environmental interference abnormality.

[0064] Step 4: Rapid Fault Location

[0065] Based on the self-diagnostic results, combined with the preset system topology and fault cause-effect graph, the fault source is accurately located, solving the problems of lagging and large errors in fault location in existing technologies; the specific operation is as follows:

[0066] 1) Basic Data Preparation: Pre-determine the topology and fault cause-effect graph of the hydropower station monitoring system. The system topology clearly defines the hierarchical relationships between the equipment layer, transmission layer, monitoring layer, and application layer, and marks the connection methods and data interaction paths between each device and module; the fault cause-effect graph is constructed based on common fault cases and equipment operation mechanisms of the hydropower station monitoring system, including the relationships between fault types, fault characteristics, fault causes, and fault impact range.

[0067] 2) Correlation Analysis and Localization: When the self-diagnosis result indicates a system anomaly, the abnormal features in the fusion feature set are extracted and matched with the feature identifiers in the fault cause-effect graph to determine the fault type. Then, based on the system topology, the associated devices and modules corresponding to the fault type are analyzed, and the failure probability of each associated device and module is calculated using the Bayesian network algorithm. The device or module with the highest failure probability is the fault source. Finally, the specific location of the fault source, the faulty component, and the fault triggering factors are output.

[0068] Step 5: Fault Level Classification and Emergency Response Coordination

[0069] This system is used to classify faults according to their severity and trigger corresponding emergency response commands to enable rapid fault handling and prevent escalation. The specific operation is as follows:

[0070] 1) Fault Level Classification: Based on the impact range of the fault source, the severity of the fault, and the speed of fault development, faults are classified into four levels: Level 1 faults are minor abnormalities of a single non-critical device that do not affect the overall operation of the system; Level 2 faults are minor abnormalities of a single critical device or multiple non-critical devices that have a slight impact on the operation of the system; Level 3 faults are abnormalities of multiple critical devices or a severe abnormality of a single core device that affect the normal operation of some functions of the system; Level 4 faults are severe abnormalities of core devices or multi-system linkage faults that may lead to system shutdown.

[0071] 2) Emergency Response Command Trigger: Based on different fault levels, corresponding emergency response commands are triggered. Level 1 faults trigger alarm notification commands; Level 2 faults trigger parameter adjustment commands and alarm notification commands; Level 3 faults trigger equipment start / stop commands, parameter adjustment commands, and operation and maintenance scheduling commands; Level 4 faults trigger emergency shutdown commands, comprehensive alarm commands, and emergency dispatch commands, enabling rapid fault handling.

[0072] Step Six: Adaptive Optimization of the Diagnostic Model

[0073] This is used to dynamically optimize the parameters of the self-diagnostic model based on diagnostic and fault handling feedback, improving the model's diagnostic accuracy and adaptability, and solving the problem that existing technical models cannot adapt to changes in system operating status; the specific operation is as follows:

[0074] 1) Optimize data collection: Collect self-diagnosis results, fault location results, actual fault information and handling feedback data during the fault handling process, with a focus on collecting cases of false alarms and missed alarms in the model and data on newly emerging fault types.

[0075] 2) Model parameter adjustment: Based on the collected optimization data, multiple parameter adjustments are made: First, the feature weight allocation coefficients are adjusted, and the weight ratio of each feature is optimized according to the deviation between the fault location result and the actual fault; second, the loss function and activation function of the model are optimized to improve the diagnostic accuracy of the model; third, training samples are supplemented, and newly emerging fault type data are added to the model training sample library to retrain the model and update the fault identification library of the model to ensure that the model can adaptively adapt to changes in the system's operating status.

[0076] Step 7: Diagnosis and Fault Log Retention

[0077] This is used to record the entire self-diagnosis and fault handling process, providing a basis for subsequent fault backtracking, model optimization, and operation and maintenance management; the specific operation is as follows:

[0078] 1) Log recording: The recorded content includes data acquisition time, preprocessing parameters, fusion feature set, self-diagnosis results, fault location information, fault level, emergency response instructions, handling results and model optimization parameters, to ensure the integrity and traceability of the log content.

[0079] 2) Log storage and retrieval: The logs are encrypted and stored using encryption algorithms. The storage period meets the traceability requirements of hydropower station operation and maintenance, and prevents log data leakage or tampering. At the same time, a log retrieval system is established to support multi-dimensional retrieval by time, fault type and equipment type, so that operation and maintenance personnel can quickly query relevant log information.

[0080] Step 8: Fault Backtracking Analysis

[0081] This is used for retrospective analysis of handled faults, summarizing fault patterns, and further optimizing fault cause-effect graphs and self-diagnostic models; the specific operation is as follows:

[0082] Regularly collect logs of handled faults, analyze the root causes, development patterns, and shortcomings in the handling process, and generate a fault retrospective report; based on the retrospective report, update the fault cause-effect graph and supplement new fault causal relationships; at the same time, incorporate the optimization suggestions obtained from the retrospective analysis into the model parameter optimization process to further improve the model's diagnostic accuracy and fault location capabilities.

[0083] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0084] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A self-diagnosis and rapid fault location method for a hydropower station monitoring system, characterized in that, Includes the following steps: Step 1: Full-dimensional data collection and preprocessing of the monitoring system. Collect the operation data, status data and environmental data of equipment at all levels of the hydropower station monitoring system. Standardize, denoise and complete the collected data to obtain standardized data. Step 2: Multimodal feature fusion and extraction. Based on the preprocessed standardized data, equipment operation features, abnormal status features, and environmental interference features are extracted respectively. Multimodal feature fusion is completed through feature weight allocation algorithm to obtain fused feature set. Step 3: Hierarchical self-diagnosis verification. Construct a lightweight self-diagnosis model at the edge and a cloud-based fusion self-diagnosis model. The edge model performs a preliminary self-diagnosis on the fusion feature set, and the cloud model performs a secondary verification on the edge diagnosis results, outputting the self-diagnosis results, which include whether the system is normal, abnormal, and the type of abnormality. Step 4: Rapid fault location. When the self-diagnosis result indicates a system anomaly, based on the preset hydropower station monitoring system topology and fault cause-effect graph, combined with the anomaly features of the fused feature set, the specific location of the fault source, the faulty component, and the fault triggering factors are located through the correlation analysis algorithm. Step 5: Fault Level Classification and Emergency Response. Based on the impact range of the fault source, the severity of the fault, and the speed of fault development, fault levels are classified, and corresponding emergency response commands are triggered based on different fault levels to achieve rapid fault handling. Step Six: Adaptive Optimization of Diagnostic Model. Based on self-diagnosis results, fault location results, and fault handling feedback data, the parameters of the lightweight self-diagnosis model at the edge and the cloud-integrated self-diagnosis model are dynamically adjusted to improve the model's diagnostic accuracy. Step 7: Diagnosis and Fault Log Retention. Record the self-diagnosis process, fault location results, fault handling process, and model optimization parameters to form a complete diagnosis and fault log for subsequent traceability and analysis.

2. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: In step one, the operating data includes real-time operating parameters of the hydropower station's turbine generator set, speed regulation system, excitation system, water conveyance system, and power distribution system. The status data includes the start / stop status of each device, fault alarm status, and component wear status. The environmental data includes temperature, humidity, vibration intensity, and electromagnetic interference intensity inside the hydropower station powerhouse. The preprocessing specifically includes: standardizing the data using a standardization algorithm, removing noise interference from the data using a denoising algorithm, and completing missing data using a data completion algorithm to ensure the integrity and accuracy of the standardized data.

3. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: In step two, the equipment operating characteristics include the fluctuation amplitude, rate of change, and steady-state deviation of the operating parameters; The abnormal status characteristics include the triggering frequency, duration, and number of associated signals of the alarm signals; The environmental interference characteristics include the deviation of environmental parameters from standard thresholds and the duration of interference. The feature weight allocation algorithm adopts the analytic hierarchy process (AHP) to allocate weights based on the degree of influence of each feature on fault diagnosis. After the weight allocation is completed, the fused feature set is obtained by weighted summation.

4. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: In step three, the lightweight self-diagnosis model at the edge adopts a pruned convolutional neural network model to achieve local real-time diagnosis. The cloud-based fusion self-diagnostic model adopts a fusion model combining Transformer and Support Vector Machine to perform secondary verification of the diagnostic results at the edge. The abnormal types of the self-diagnostic results include equipment hardware failure, software program failure, data transmission failure, and environmental interference abnormalities.

5. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: In step four, the preset hydropower station monitoring system topology includes the hierarchical relationship between the equipment layer, transmission layer, monitoring layer and application layer, and clarifies the connection method and data interaction path between each device and module; The fault cause-effect graph is constructed based on common fault cases and equipment operation mechanisms of hydropower station monitoring systems, and includes the correlation between fault types, fault characteristics, fault causes and fault impact range. The correlation analysis algorithm employs a Bayesian network algorithm, which achieves accurate location of the fault source by matching the abnormal features of the fused feature set with the fault causal graph.

6. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: In step five, the fault levels are divided into four levels: Level 1 fault is a minor abnormality of a single non-critical device that does not affect the overall operation of the system; Level 2 fault is a minor abnormality of a single critical device or multiple non-critical devices that have a minor impact on the operation of the system; Level 3 fault is multiple critical devices or a single core device that has a serious abnormality that affects the normal operation of some functions of the system; Level 4 fault is a serious abnormality of a core device or a multi-system linkage fault that may lead to system shutdown. The emergency response commands include equipment start / stop commands, parameter adjustment commands, alarm notification commands, and operation and maintenance scheduling commands. Different fault levels correspond to different emergency response commands.

7. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: In step six, the dynamic adjustment of model parameters specifically includes: adjusting the feature weight allocation coefficients based on the deviation between the fault location results and the actual fault; optimizing the model's loss function and activation function based on diagnostic false alarms and missed alarms; and supplementing training samples and updating the model's fault identification library based on newly emerging fault types to ensure that the model adaptively adapts to changes in the system's operating state.

8. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: In step seven, the diagnostic and fault log includes data acquisition time, preprocessing parameters, fusion feature set, self-diagnosis results, fault location information, fault level, emergency response instructions, handling results, and model optimization parameters. The log is stored in an encrypted manner, and the storage period meets the traceability requirements of hydropower station operation and maintenance. It supports retrieval by time, fault type, and equipment type.

9. The self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: The method also includes a fault backtracking analysis step, which, based on diagnosis and fault logs, backtracks the handled faults, analyzes the root causes, development patterns, and shortcomings in the handling process of the faults, and generates a fault backtracking report to optimize the fault cause-effect graph and self-diagnosis model.

10. A self-diagnosis and rapid fault location method for a hydropower station monitoring system according to claim 1, characterized in that: The method is applicable to monitoring systems of hydropower stations of different sizes. By adjusting the data acquisition range, feature dimensions and model parameters, it can be adapted to the monitoring needs of small, medium and large hydropower stations. The adaptation process does not require modification of the monitoring system hardware.

Images