Fault recording and intelligent diagnosis method for pumped storage hydropower station

Abstract

The invention relates to a fault recording and intelligent diagnosis method for a pumped storage hydropower station. The method is suitable for the technical field of water conservancy and hydropower engineering. According to the technical scheme, the method comprises the following steps: receiving operation condition parameters of a hydropower station, determining a condition identifier based on the operation condition parameters, and respectively determining an electrical quantity sampling frequency and a mechanical quantity sampling frequency based on the condition identifier; based on the electrical quantity sampling frequency and the mechanical quantity sampling frequency, recording data corresponding to the working condition identification are obtained, and the recording data comprise electrical quantity sampling signals and mechanical quantity sampling signals of the hydropower station; feature extraction and fusion are carried out on the recording data corresponding to the working condition identification, a trained basic diagnosis model is input, and a diagnosis result including the probability of various preset faults is output; if a fault with the probability greater than a preset value exists in the probabilities of various preset faults, outputting a diagnosis result; on the contrary, based on hydropower station operation condition parameters, a fault rule base of the auxiliary diagnosis model is reasoned according to rules, and the diagnosis result is corrected.

Description

Fault Recording and Intelligent Diagnosis Methods for Pumped Storage Hydropower Stations Technical Field

[0001] This invention relates to a fault recording and intelligent diagnosis method for pumped storage hydropower stations. It is applicable to the field of water conservancy and hydropower engineering. Background Technology

[0002] Pumped-storage hydropower stations are core peak-shaving facilities for power grids. Due to their bidirectional operation of pumping and power generation, frequent switching of operating conditions, and diverse and highly coupled equipment fault types, monitoring and diagnosis are extremely difficult. Existing technologies have four major shortcomings: ① Poor specificity of waveform recording, with general-purpose devices failing to capture key signals such as transients during operating condition switching and pump start-up impacts; ② Simple data processing, resulting in large redundancy of raw data and low diagnostic efficiency; ③ Weak adaptability of diagnostic models, with single algorithms struggling to identify coupled faults, leading to low accuracy and high false positive rates; ④ Lack of early warning and rapid source tracing capabilities, resulting in delayed fault handling. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a fault recording and intelligent diagnosis method for pumped storage hydropower stations, addressing the aforementioned problems.

[0004] The technical solution adopted in this invention is: a fault recording and intelligent diagnosis method for pumped storage hydropower stations, comprising: receiving operating condition parameters of the hydropower station; determining operating condition identifiers based on the operating condition parameters; and determining electrical quantity sampling frequencies and mechanical quantity sampling frequencies based on the operating condition identifiers; acquiring recording data corresponding to the operating condition identifiers based on the electrical quantity sampling frequencies and mechanical quantity sampling frequencies, including electrical quantity sampling signals and mechanical quantity sampling signals of the hydropower station; extracting and fusing features from the recording data corresponding to the operating condition identifiers, inputting them into a trained basic diagnostic model, and outputting diagnostic results, including the probabilities of various preset faults; if there is a fault with a probability greater than a preset value among the probabilities of various preset faults, then outputting a diagnostic result; otherwise, based on the operating condition parameters of the hydropower station, correcting the diagnostic result by referring to the fault rule base of the rule-based reasoning auxiliary diagnostic model.

[0005] It also includes: starting with the faulty equipment in the diagnostic results, combining the operating condition parameters, traversing the fault association map, locating the root cause of the fault and the scope of its impact; the fault association map is constructed based on the topological relationship of hydropower station equipment and historical fault propagation paths, and includes equipment nodes and association relationships.

[0006] The determination of operating condition identifiers based on operating condition parameters includes: using speed, load, and switching commands as indicators to determine corresponding categories of operating condition identifiers, including stable power generation operating condition, stable power generation operating condition, stable pumping operating condition, stable pumping operating condition, and transitional operating condition switching condition.

[0007] Based on the operating condition identifier, the sampling frequencies for electrical quantities and mechanical quantities are determined respectively, including: Stable power generation operating condition: electrical quantity sampling frequency 5kHz, mechanical quantity sampling frequency 2kHz, storage duration 10 minutes / segment; Power generation start-up operating condition: electrical quantity sampling frequency 20kHz, mechanical quantity sampling frequency 10kHz, storage duration from the start-up command to 5 minutes after the speed stabilizes; Stable pumping operating condition: electrical quantity sampling frequency 5kHz, mechanical quantity sampling frequency 2kHz, storage duration 10 minutes / segment; Pumping start-up operating condition: electrical quantity sampling frequency 20kHz, mechanical quantity sampling frequency 10kHz, storage duration from the start-up command to 5 minutes after the speed stabilizes; Transitional operating condition: electrical quantity sampling frequency 20kHz, mechanical quantity sampling frequency 10kHz, storage duration from the switching command to 5 minutes after the operating condition switch is completed.

[0008] The process of extracting and fusing features from the waveform data corresponding to the operating condition identifier includes: removing outliers from the collected raw data using the 3σ criterion; aligning the multi-channel data to time based on GPS timestamps; extracting features from electrical quantity sampling signals and mechanical quantity sampling signals using a combination of time-domain analysis, frequency-domain analysis, and time-frequency-domain analysis; and assigning weights to the extracted features using an attention mechanism.

[0009] The basic diagnostic model includes a CNN network and an LSTM network. The CNN network contains convolutional layers and attention layers, and the LSTM network contains hidden layers. The output layer uses the Softmax function to output the probability of various preset faults.

[0010] The preset value is 0.85.

[0011] A fault recording and intelligent diagnosis device for pumped storage hydropower stations includes: a sampling frequency determination module, used to receive operating condition parameters of the hydropower station, determine operating condition identifiers based on the operating condition parameters, and determine electrical quantity sampling frequencies and mechanical quantity sampling frequencies based on the operating condition identifiers; a signal sampling module, used to acquire recording data corresponding to the operating condition identifiers based on the electrical quantity sampling frequencies and mechanical quantity sampling frequencies, including electrical quantity sampling signals and mechanical quantity sampling signals of the hydropower station; a model diagnosis module, used to extract and fuse features from the recording data corresponding to the operating condition identifiers, input them into a trained basic diagnosis model, and output diagnosis results, including the probabilities of various preset faults; and a result output module, used to output diagnosis results if there is a fault with a probability greater than a preset value among the probabilities of various preset faults; otherwise, based on the operating condition parameters of the hydropower station, the diagnosis results are corrected by referring to the fault rule base of the rule-based reasoning auxiliary diagnosis model.

[0012] A storage medium storing a computer program executable by a processor, wherein the computer program, when executed, implements the steps of the claimed pumped-storage hydropower station fault recording and intelligent diagnosis method.

[0013] A fault recording and intelligent diagnosis device for pumped storage hydropower stations includes a memory and a processor. The memory stores a computer program that can be executed by the processor. When the computer program is executed, it implements the steps of the fault recording and intelligent diagnosis method for pumped storage hydropower stations.

[0014] The beneficial effects of this invention are: based on the operating condition identifier, this invention determines the corresponding electrical quantity sampling signal and mechanical quantity sampling signal of the hydropower station, realizes adaptive waveform recording based on operating condition, and improves the signal integrity during the switching phase.

[0015] This invention reduces data redundancy by preprocessing the waveform data through cleaning, synchronous calibration, and multi-dimensional feature extraction, providing high-quality feature data for subsequent diagnosis and improving diagnostic efficiency.

[0016] This invention employs a diagnostic approach that integrates a basic diagnostic model with a rule-based reasoning-assisted diagnostic model. This approach utilizes deep learning models to capture the nonlinear characteristics and temporal correlations of complex faults, while also correcting fuzzy diagnostic results through rule-based reasoning. This improves the diagnostic accuracy for multi-device coupled faults and complex operating conditions, and reduces the false positive rate.

[0017] This invention rapidly locates the root cause and scope of a fault based on a fault correlation graph. The early warning module provides tiered early warnings, offering maintenance personnel precise fault handling guidelines, shortening fault handling time, and improving the operational stability of hydropower stations. Attached Figure Description

[0018] Figure 1 is a flowchart of an embodiment. Detailed Implementation

[0019] The embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0020] In the description of this invention, "multiple" means two or more. The use of "first" and "second" is for distinguishing technical features only and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or the order of the indicated technical features. Furthermore, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art.

[0021] As shown in Figure 1, this embodiment is a fault recording and intelligent diagnosis method for pumped storage hydropower stations, which specifically includes the following steps: S100, receiving the operating condition parameters of the pumped storage hydropower station, determining the operating condition identifier based on the operating condition parameters, and determining the sampling frequency of electrical quantities and mechanical quantities based on the operating condition identifier.

[0022] In this embodiment, the operating parameters of the hydropower station are collected by industrial-grade sensors (1Hz-100kHz) and deployed in key parts of the hydropower station.

[0023] In this example, operating condition identification uses speed, load, and switching commands as indicators to output specific operating condition identifier types. Operating condition identification employs a fuzzy comprehensive evaluation method, selecting speed (≥250 r / min as the power generation start-up threshold, ≤100 r / min as the pumping start-up threshold), load (≥50 MW as the power generation stability threshold, ≤-50 MW as the pumping stability threshold), and operating condition switching commands (from the hydropower station monitoring system) as evaluation indicators. The operating conditions are divided into five categories: stable power generation, power generation start-up, stable pumping operation, pumping start-up, and transitional operating conditions, with real-time output of operating condition identifier signals.

[0024] This embodiment determines the sampling frequencies of electrical and mechanical quantities based on the operating condition identifier and the waveform recording parameter configuration library. The configuration library presets waveform recording parameters for different operating conditions: Stable power generation condition: Electrical quantities (voltage, current, power) sampling frequency 5kHz, mechanical quantities (generator vibration, temperature, turbine pressure, flow rate) sampling frequency 2kHz, storage duration 10 minutes / segment; Power generation start-up condition: Electrical quantities sampling frequency 20kHz, mechanical quantities sampling frequency 10kHz, storage duration from the start-up command to 5 minutes after speed stabilization; Stable pumping condition: Electrical quantities sampling frequency 5kHz, mechanical quantities (pump vibration, temperature, inlet and outlet pressure, flow rate) sampling frequency 2kHz, storage duration 10 minutes / segment; Pumping start-up condition: Electrical quantities sampling frequency 20kHz, mechanical quantities sampling frequency 10kHz, storage duration from the start-up command to 5 minutes after speed stabilization; Transitional operating condition: Electrical quantities sampling frequency 20kHz, mechanical quantities sampling frequency 10kHz, storage duration from the switching command to 5 minutes after the operating condition switch is completed.

[0025] S200: Based on the electrical quantity sampling frequency and the mechanical quantity sampling frequency, acquire the waveform data corresponding to the operating condition identifier, including the electrical quantity sampling signal and the mechanical quantity sampling signal of the hydropower station.

[0026] S300: Extract and fuse features from the waveform data corresponding to the operating condition identifier, input the data into the trained basic diagnostic model, and output the diagnostic results, including the probability of various preset faults.

[0027] S310. Perform data preprocessing on the sampling signal data corresponding to the operating condition identifier, including data cleaning, data synchronization calibration, and feature extraction.

[0028] Data cleaning: Outlier removal is performed on the collected raw data using the 3σ criterion. The mean μ and standard deviation σ of the data are calculated, and data exceeding the range of [μ-3σ, μ+3σ] are removed. Missing data is supplemented using linear interpolation. Data synchronization calibration: Multi-channel data is time-aligned based on GPS timestamps to correct transmission delays from different sensors and ensure that multi-dimensional data at the same time correspond and match. Feature extraction: Features are extracted using a combination of time-domain analysis, frequency-domain analysis, and time-frequency-domain analysis. Time-domain features include voltage / current peak value, mean, variance, kurtosis, and temperature change rate. Frequency-domain features are extracted using FFT transform to obtain the dominant frequency and harmonic components (2nd-5th harmonics). Time-frequency-domain features are extracted using wavelet transform to obtain wavelet entropy and short-time Fourier transform peak values. Finally, a 28-dimensional fused feature vector is output.

[0029] S320, Feature Fusion. An attention mechanism is used to assign weights to the 28-dimensional feature vector. Through training, the weight coefficients of fault-related features are obtained (such as the weight coefficient of generator vibration frequency is 0.15, and the weight coefficient of transformer winding temperature change rate is 0.12), thereby enhancing key features.

[0030] S330. Input the fused features into the trained basic diagnostic model and output the diagnostic results.

[0031] In this embodiment, the basic diagnostic model includes a CNN network and an LSTM network. The CNN network contains three convolutional layers (with kernel sizes of 3×1, 5×1, and 7×1) and one attention layer. The LSTM network contains two hidden layers (with 64 and 32 hidden units respectively). The output layer uses the Softmax function to output the probability of 12 preset fault types (including generator stator grounding fault, turbine cavitation fault, transformer inter-turn short circuit fault, etc.).

[0032] S400. If there is a fault with a probability greater than the preset value (e.g., 0.85) among the various preset fault probabilities, then output the diagnosis result; otherwise, based on the hydropower station operating condition parameters, compare the fault rule base of the rule reasoning-assisted diagnosis model, perform rule matching, and correct the diagnosis result.

[0033] In this embodiment, the fault rule base of the rule-based reasoning-assisted diagnosis model is built based on 500+ historical fault cases and the experience of 10 experts, including 200+ rules such as "generator vibration main frequency ≥ 50Hz and current peak value ≥ 1.2 times the rated value → generator rotor imbalance fault".

[0034] S500, based on diagnostic results and operating condition parameters, combined with fault correlation diagrams, traces the source of faults.

[0035] In this embodiment, the fault association map is constructed based on the topological relationships of hydropower station equipment (such as generator → transformer → busbar → power grid, turbine → speed control system → hydraulic device) and historical fault propagation paths, including 120+ equipment nodes and 300+ association relationships.

[0036] In this example, the source tracing algorithm uses a depth-first search method. Starting from the faulty equipment in the diagnostic results, it traverses the correlation graph in combination with real-time operating parameters (such as load and speed when the fault occurs) to locate the root cause of the fault (such as tracing the fault to rotor imbalance or bearing wear when the fault is diagnosed as generator vibration fault) and the scope of influence (such as the scope of influence of rotor imbalance fault is generator → transformer → grid frequency).

[0037] In some specific embodiments, thresholds are set according to the fault level (minor fault, general fault, severe fault) and the degree of deviation of characteristic parameters: Level 1 warning (minor fault): characteristic parameters exceed the normal range by 10%-20%, such as transformer oil temperature exceeding the normal range by 15%; Level 2 warning (general fault): characteristic parameters exceed the normal range by 20%-50%, such as generator vibration peak exceeding the normal range by 30%; Level 3 warning (severe fault): characteristic parameters exceed the normal range by ≥50%, such as stator current exceeding the rated value by 1.5 times; warning signals are output through audible and visual alarms (visual interactive module) and SMS / APP push (mobile terminal of maintenance personnel).

[0038] In some specific embodiments, a three-dimensional visualization display is implemented based on WebGL. The real-time display content includes: operating condition panel (operating condition type, speed, load, etc.), waveform data curves (real-time curves and historical curves of voltage, current, vibration, etc.), characteristic parameter heat map, fault diagnosis results (fault type, level, time), and three-dimensional map of fault source path; and provides data query (query by operating condition, time, fault type), diagnostic result export (PDF / Excel format), and fault alarm history record query functions.

[0039] The following is a specific example: During a 12-month trial operation of a pumped storage hydropower station (1200MW installed capacity, 4 units of 300MW each), 32 artificially simulated fault tests were conducted (covering newly added fault types such as pump cavitation and excitation system demagnetization). Seven real faults were captured, and the overall system performance was excellent: the fault recording integrity rate was 100%, and compared with traditional recording devices, the integrity of transient signal acquisition during the operating condition switching phase was improved by 85%. The capture delay of the pump start-up impact signal was shortened from 20ms in traditional devices to 1ms. The fault diagnosis accuracy rate was 96%, of which the fault diagnosis accuracy rate for a single device reached 98%. The system achieved a 5% accuracy rate in diagnosing multi-device coupled faults, reaching 92%, a 22% improvement compared to a single CNN model. The false positive rate was reduced to 2%, with no missed diagnoses. The average fault tracing time was ≤30 seconds, a 98.9% reduction compared to manual tracing (average 4.5 hours), and it could accurately pinpoint specific components (e.g., generator rotor imbalance could be located to a specific rotor blade). The early warning module achieved a 92% accuracy rate for minor faults, with an early warning lead time of up to 25 minutes for water pump cavitation faults, providing ample time for maintenance personnel to handle the situation. During the trial operation, downtime due to faults decreased by 68% compared to the same period last year, directly reducing economic losses by approximately 12 million yuan. Furthermore, the system has a self-learning function. During the trial operation, after optimizing model parameters using 15 newly added fault data cases, the diagnostic accuracy further improved to 96.5%, fully validating the system's stability and scalability. Trial operation verification: In 25 faults (5 real ones), the waveform integrity rate was 100%, the diagnostic accuracy rate was 96%, the source tracing time was ≤30s, the early warning accuracy rate was 92%, and the downtime was reduced by 68%.

[0040] Example 2: This example describes a fault recording and intelligent diagnosis device for a pumped storage hydropower station, specifically including: a sampling frequency determination module, used to receive the operating condition parameters of the hydropower station, determine the operating condition identifier based on the operating condition parameters, and determine the sampling frequencies of electrical quantities and mechanical quantities based on the operating condition identifier; a signal sampling module, used to acquire the recording data corresponding to the operating condition identifier based on the sampling frequencies of electrical quantities and mechanical quantities, including the sampling signals of electrical quantities and mechanical quantities of the hydropower station; a model diagnosis module, used to extract and fuse features from the recording data corresponding to the operating condition identifier, input it into a trained basic diagnosis model, and output the diagnosis result, including the probability of various preset faults; a result output module, used to output the diagnosis result if there is a fault with a probability greater than a preset value among the probabilities of various preset faults; otherwise, based on the operating condition parameters of the hydropower station, the diagnosis result is corrected by referring to the fault rule base of the rule reasoning auxiliary diagnosis model.

[0041] Example 3: This example is a storage medium that stores a computer program that can be executed by a processor. When the computer program is executed, it implements the steps of the pumped storage hydropower station fault recording and intelligent diagnosis method described in Example 1.

[0042] Example 4: This example is a fault recording and intelligent diagnosis device for pumped storage hydropower stations, which has a memory and a processor. The memory stores a computer program that can be executed by the processor. When the computer program is executed, it implements the steps of the fault recording and intelligent diagnosis method for pumped storage hydropower stations described in Example 1.

[0043] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0044] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0045] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the aforementioned program can be printed, because the aforementioned program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0046] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0047] In the foregoing description of this specification, references to terms such as "one embodiment," "another embodiment," or "some embodiments" indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0048] Although embodiments of the invention have been shown and described, those skilled in the art will understand 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 claims and their equivalents.

[0049] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for fault recording and intelligent diagnosis of pumped storage hydropower stations, characterized in that, include: Receive the operating condition parameters of the hydropower station, determine the operating condition identifier based on the operating condition parameters, and determine the sampling frequency of electrical quantities and mechanical quantities based on the operating condition identifier; Based on the sampling frequencies of electrical and mechanical quantities, waveform data corresponding to the operating condition identifiers are acquired, including electrical and mechanical sampling signals of the hydropower station. Features are extracted and fused from the waveform data corresponding to the operating condition identifiers, and input into the trained basic diagnostic model to output diagnostic results, including the probabilities of various preset faults. If there is a fault with a probability greater than the preset value among the probabilities of various preset faults, a diagnostic result is output. Otherwise, based on the operating condition parameters of the hydropower station, the diagnostic result is corrected by referring to the fault rule base of the rule-based reasoning auxiliary diagnostic model.

2. The method for fault recording and intelligent diagnosis of pumped storage hydropower stations according to claim 1, characterized in that, Also includes: Starting with the faulty equipment in the diagnostic results, and combining the operating condition parameters, the fault correlation graph is traversed to locate the root cause of the fault and its scope of influence. The fault association map is constructed based on the topological relationship of hydropower station equipment and historical fault propagation paths, and includes equipment nodes and association relationships.

3. The method for fault recording and intelligent diagnosis of pumped storage hydropower stations according to claim 1, characterized in that, The determination of operating condition identifiers based on operating condition parameters includes: using speed, load, and switching commands as indicators to determine corresponding categories of operating condition identifiers, including stable power generation operating condition, stable power generation operating condition, stable pumping operating condition, stable pumping operating condition, and transitional operating condition switching condition.

4. The method for fault recording and intelligent diagnosis of pumped storage hydropower stations according to claim 1, characterized in that, Based on the operating condition identifier, the sampling frequencies for electrical quantities and mechanical quantities are determined respectively, including: Stable power generation operating condition: electrical quantity sampling frequency 5kHz, mechanical quantity sampling frequency 2kHz, storage duration 10 minutes / segment; Power generation start-up operating condition: electrical quantity sampling frequency 20kHz, mechanical quantity sampling frequency 10kHz, storage duration from the start-up command to 5 minutes after the speed stabilizes; Stable pumping operating condition: electrical quantity sampling frequency 5kHz, mechanical quantity sampling frequency 2kHz, storage duration 10 minutes / segment; Pumping start-up operating condition: electrical quantity sampling frequency 20kHz, mechanical quantity sampling frequency 10kHz, storage duration from the start-up command to 5 minutes after the speed stabilizes; Transitional operating condition: electrical quantity sampling frequency 20kHz, mechanical quantity sampling frequency 10kHz, storage duration from the switching command to 5 minutes after the operating condition switch is completed.

5. The method for fault recording and intelligent diagnosis of pumped storage hydropower stations according to claim 1, characterized in that, The process of extracting and fusing features from the waveform data corresponding to the operating condition identifier includes: removing outliers from the collected raw data using the 3σ criterion; aligning the multi-channel data to time based on GPS timestamps; extracting features from electrical quantity sampling signals and mechanical quantity sampling signals using a combination of time-domain analysis, frequency-domain analysis, and time-frequency-domain analysis; and assigning weights to the extracted features using an attention mechanism.

6. The method for fault recording and intelligent diagnosis of pumped storage hydropower stations according to claim 1, characterized in that, The basic diagnostic model includes a CNN network and an LSTM network. The CNN network contains convolutional layers and attention layers, and the LSTM network contains hidden layers. The output layer uses the Softmax function to output the probability of various preset faults.

7. The method for fault recording and intelligent diagnosis of pumped storage hydropower stations according to claim 1, characterized in that, The preset value is 0.

85.

8. A fault recording and intelligent diagnostic device for pumped storage hydropower stations, characterized in that, include: The sampling frequency determination module is used to receive the operating condition parameters of the hydropower station, determine the operating condition identifier based on the operating condition parameters, and determine the sampling frequency of electrical quantities and mechanical quantities based on the operating condition identifier. The signal sampling module is used to acquire waveform data corresponding to the operating condition identifier based on the electrical quantity sampling frequency and the mechanical quantity sampling frequency, including electrical quantity sampling signals and mechanical quantity sampling signals of hydropower stations; The model diagnostic module is used to extract and fuse features from the waveform data corresponding to the working condition identifier, input the trained basic diagnostic model, and output the diagnostic results, including the probability of various preset faults. The result output module is used to output the diagnostic result if there is a fault with a probability greater than the preset value among the various preset fault probabilities. Conversely, based on the operating parameters of the hydropower station, the fault rule base of the rule-based reasoning-assisted diagnostic model is used to correct the diagnostic results.

9. A storage medium storing a computer program executable by a processor, characterized in that, When the computer program is executed, it implements the steps of the pumped storage hydropower station fault recording and intelligent diagnosis method according to any one of claims 1 to 7.

10. A fault recording and intelligent diagnostic device for a pumped-storage hydropower station, comprising a memory and a processor, wherein the memory stores a computer program executable by the processor, characterized in that, When the computer program is executed, it implements the steps of the pumped storage hydropower station fault recording and intelligent diagnosis method according to any one of claims 1 to 7.

Images