A ground net corrosion fault diagnosis method, system, storage medium and equipment

Abstract

The present application relates to the technical field of power system operation and maintenance, and discloses a grounding net corrosion fault diagnosis method and system, a storage medium and equipment, which comprises the following steps: multiple types of sensors are arranged at key points of the grounding net to collect electrical signals and environmental parameter data of the grounding net in real time, and the collected original data is preprocessed to generate a high-quality multi-source data set; multi-source data is subjected to feature extraction, and the extracted multi-source features are normalized and weightedly fused to generate a comprehensive fault feature vector; a diagnosis model suitable for multi-feature input is constructed, a whale optimization algorithm (WOA) is used to adaptively optimize parameters and feature fusion weights of the diagnosis model, the diagnosis model is trained using the comprehensive fault feature vector, and the model is optimized; the comprehensive fault feature vector is input into the optimized diagnosis model, corrosion fault diagnosis results are output, a fault report and alarm information are generated, and accurate identification, positioning and early warning of grounding net corrosion are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system operation and maintenance, and in particular to a grounding grid corrosion fault diagnosis method and system, a storage medium and equipment. BACKGROUND

[0002] In the power system, the grounding grid is the cornerstone of ensuring the safe operation of key facilities such as substations and power plants, and undertakes the core responsibilities of discharging fault current, stabilizing equipment potential and ensuring personnel safety. The integrity of its conductor network is directly related to the reliability of the power grid. However, the grounding grid is buried in complex underground soil environment for a long time, facing many challenges such as electrochemical corrosion and stray current corrosion, resulting in a decrease in the effective cross-sectional area of the conductor, the deterioration of the connection point, and further causing serious problems such as abnormal increase of grounding resistance and uneven potential distribution. Especially in the context of rapid development of high-voltage, ultra-high-voltage and new energy stations, the scale of the grounding grid is expanding, and the operating environment is becoming increasingly complex, making the risk of corrosion failure increasing day by day.

[0003] At present, two traditional methods are mainly relied on in engineering practice for grounding grid state assessment: one is regular manual inspection, which judges the corrosion condition through local excavation and visual inspection; the other is periodic grounding resistance testing with circuit breaker cooperation. These methods are widely used, but have significant defects. They are all offline and non-continuous detection methods, with long detection period and poor real-time performance, and cannot timely discover early or developing corrosion hazards. More importantly, the grounding resistance is a macro-lagging parameter, and when it changes significantly, the corrosion is usually quite serious, missing the best maintenance opportunity.

[0004] In recent years, the Internet of Things technology has promoted the application of online monitoring, such as continuous measurement of grounding downlead current or ground potential difference. However, most of these methods rely on a single electrical parameter, and have limited diagnostic capability. Grounding grid corrosion is a complex process of multi-factor coupling, which is deeply affected by environmental factors such as soil temperature and humidity in addition to electrical load. The single parameter method is extremely sensitive to environmental interference and background noise, and it is difficult to effectively extract weak early corrosion features, resulting in high false positive and false negative rates.

[0005] Therefore, there is an urgent need in the art for a new type of grounding grid corrosion fault diagnosis method that can deeply integrate multi-source information and has intelligent self-adaptive optimization capability, to realize accurate identification and early warning of early corrosion and improve the intelligent level and safety and reliability of power grid operation and maintenance. SUMMARY

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a grounding grid corrosion fault diagnosis method based on multi-source sensor information fusion and whale optimization algorithm. By feature extraction and adaptive weight fusion, combined with intelligent algorithm optimization of the diagnostic model, the method can achieve accurate identification, location and early warning of grounding grid corrosion.

[0007] On the one hand, a method for diagnosing grounding grid corrosion faults is provided, including the following steps:

[0008] S1: By deploying various types of sensors at key points of the grounding grid, the electrical signals and environmental parameters of the grounding grid are collected in real time. The collected raw data is then processed for time synchronization, filtering, noise reduction, and standardization to generate a high-quality multi-source dataset.

[0009] S2: Perform feature extraction on the processed multi-source data, including time-domain and frequency-domain analysis of electrical signals to extract electrical features, statistical calculation of environmental parameters to extract environmental features, and normalize and weighted fuse the extracted multi-source features to generate a comprehensive fault feature vector;

[0010] S3: Construct a diagnostic model suitable for multiple feature inputs, use the Whale Optimization Algorithm (WOA) to adaptively optimize the parameters and feature fusion weights of the diagnostic model, and use the fused comprehensive fault feature vector to train the diagnostic model to obtain the optimized diagnostic model.

[0011] S4: Input the comprehensive fault feature vector into the optimized diagnostic model, output the corrosion fault diagnosis result, and generate a fault report and alarm information based on the diagnosis result.

[0012] Further, in step S2, the feature extraction specifically includes:

[0013] S21: The electrical features of the electrical signal, including transient amplitude features and spectral energy features, are extracted by wavelet transform or adaptive filtering. At the same time, the mean and variance of the environmental parameters, such as temperature and humidity, are calculated to obtain the environmental features.

[0014] S22: Calculate the cross-correlation coefficient between electrical signals and environmental parameters. This is used to reflect the correlation between electrical signals and environmental parameters, and the calculation formula is expressed as:

[0015] ;

[0016] in, and These are sampled values ​​of electrical signals and environmental parameters, respectively. and These are their average values.

[0017] Furthermore, in step S2, the normalization and weighted fusion of the extracted multi-source features include:

[0018] Electrical and environmental characteristics are normalized separately to unify their dimensions. Then, the characteristics are linearly weighted and fused according to their importance using linear weighting or principal component analysis to generate a comprehensive fault feature vector. The formula for the linear weighted fusion is:

[0019] ;

[0020] in, For electrical signal feature vectors, For environmental parameter feature vectors, To merge weights, and The Whale Optimization Algorithm (WOA) is adaptively determined.

[0021] Further, in step S3, the diagnostic model is a multilayer feedforward neural network or a support vector machine, and the adaptive optimization of the diagnostic model specifically includes:

[0022] The Whale Optimization Algorithm (WOA) optimizes the parameters and / or feature fusion weights of the diagnostic model through iterative optimization. In each iteration, it optimizes the parameters based on the globally optimal parameter vector of the current iteration. Update the entire parameter vector population, where each parameter vector... The update formula is:

[0023] ;

[0024] in, This represents the magnitude of the difference between the current parameter vector and the globally optimal parameter vector. and To control the coefficients of the parameter vector update direction and step size, , , and A random number in the range [0,1]. It is the convergence factor, which decreases linearly with the number of iterations.

[0025] Preferably, in step S3, a supervised learning mechanism is used to train the diagnostic model, specifically including:

[0026] The fused integrated fault feature vector As input, the corresponding corrosion fault labels are used as the expected output to construct a training sample set, and the cross-entropy loss function is used as the optimization objective for model training.

[0027] During training, an adaptive moment estimation optimizer is used in conjunction with a cosine annealing strategy to dynamically adjust the learning rate in order to accelerate model convergence and avoid getting trapped in local optima. At the same time, an L2 regularization term is introduced into the loss function of the diagnostic model to constrain the model weights and prevent overfitting. An early stopping strategy is used to automatically terminate training when the performance on the validation set no longer improves.

[0028] During training, the generalization performance of the diagnostic model was evaluated using the K-fold cross-validation method, and model selection was based on a comprehensive index of F1 score and accuracy on the validation set.

[0029] After training is complete, the final diagnostic model parameters, feature fusion weights, and mean and standard deviation parameters required for data preprocessing are saved to form a complete and deployable diagnostic model.

[0030] Furthermore, in step S4, the output corrosion fault diagnosis result specifically includes:

[0031] Define corrosion index The degree of corrosion is quantified as follows:

[0032] ;

[0033] in, This refers to the grounding resistance under normal conditions. For the current measurement of grounding resistance;

[0034] According to the corrosion index The system combines the model output results to determine whether a corrosion fault has occurred and the degree of corrosion. If the diagnostic indicators exceed the preset threshold, it is determined to be a corrosion fault and the fault location is located.

[0035] Furthermore, in step S1, the various types of sensors include one or more of a ground current sensor, a voltage sensor, a temperature sensor, and a humidity sensor;

[0036] The time synchronization is achieved by combining hardware clock synchronization and software alignment. Each data acquisition node is equipped with a high-precision clock chip to provide a unified time reference for the entire monitoring system.

[0037] The filtering and denoising are performed using a multi-stage filtering strategy, wherein,

[0038] For electrical signals, a moving average filter is used to remove obvious gross errors and impulse noise. Then, a threshold denoising method based on wavelet transform is adopted. An appropriate wavelet basis is selected for multi-scale decomposition. The high-frequency detail coefficients after decomposition are subjected to soft thresholding to effectively suppress white noise and preserve the abrupt change characteristics of the signal.

[0039] For environmental parameter data, a low-pass filter is used to smooth the data, remove high-frequency measurement noise, and retain the true trend of environmental changes.

[0040] The standardization process is performed using Z-score standardization to transform the original data into a standard normal distribution.

[0041] Preferably, in step S4, generating the fault report and alarm information includes:

[0042] It automatically generates diagnostic reports, uploads fault type, location, and severity information to the monitoring platform via the communication module, triggers on-site alarm devices, and provides a visual interface to display diagnostic results and historical trends in real time, supporting maintenance personnel in developing repair plans.

[0043] On the other hand, a grounding grid corrosion fault diagnosis system is provided, including:

[0044] The data acquisition and preprocessing module is used to collect electrical signals and environmental parameter data of the grounding grid in real time by deploying various types of sensors at key points of the grounding grid, and to perform time synchronization, filtering and noise reduction and standardization processing on the collected raw data to generate high-quality multi-source datasets.

[0045] The multi-source feature extraction and fusion module is used to extract features from the processed multi-source data, including time-domain and frequency-domain analysis of electrical signals to extract electrical features, statistical calculation of environmental parameters to extract environmental features, and normalization and weighted fusion of the extracted multi-source features to generate a comprehensive fault feature vector.

[0046] The diagnostic model construction and optimization module is used to construct a diagnostic model suitable for multiple feature inputs. The Whale Optimization Algorithm (WOA) is used to adaptively optimize the parameters and feature fusion weights of the diagnostic model. The fused comprehensive fault feature vector is used to train the diagnostic model to obtain the optimized diagnostic model.

[0047] The fault diagnosis and result output module is used to input the comprehensive fault feature vector into the optimized diagnostic model, output the corrosion fault diagnosis result, and generate a fault report and alarm information based on the diagnosis result.

[0048] In addition, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the grounding grid corrosion fault diagnosis method described in any of the above claims.

[0049] Meanwhile, an electronic device is provided, comprising: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the grounding grid corrosion fault diagnosis method described above.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] This invention, by constructing a distributed monitoring network and integrating multi-source heterogeneous electrical and environmental data, can comprehensively capture information reflecting the corrosion status of the grounding grid. Combined with the specific enhancement of weak corrosion characteristic signals, it significantly improves the sensitivity and diagnostic accuracy of early and latent corrosion faults, and can achieve precise location of fault points.

[0052] This invention employs the Whale Optimization Algorithm (WOA) to collaboratively optimize diagnostic model parameters and feature fusion weights. This overcomes the shortcomings of traditional methods that rely on experience-based trial and error or grid search, which are inefficient and prone to getting trapped in local optima. It automatically finds the global optimum, making the model more adaptable and robust in complex and changing field environments.

[0053] This invention constructs a complete automated fault diagnosis and decision support process, from real-time data acquisition and intelligent analysis and diagnosis to automatic report generation and early warning. It greatly reduces the reliance on human experience, improves operation and maintenance efficiency, and promotes the intelligent transformation and upgrading of grounding grid maintenance from "periodic maintenance" to "predictive maintenance". Attached Figure Description

[0054] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0055] Figure 1 This is a flowchart of a grounding grid corrosion fault diagnosis method according to the present invention;

[0056] Figure 2 This is a structural block diagram of a grounding grid corrosion fault diagnosis system according to the present invention;

[0057] Figure 3 This is a schematic diagram of an embodiment of an electronic device according to the present invention. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0059] This invention collects electrical and environmental data in real time by deploying a multi-source sensor network, extracts time-frequency domain features and statistical features after preprocessing, adopts the whale optimization algorithm to adaptively determine weights, and fuses multi-source features to generate a comprehensive fault feature vector. It also uses optimization algorithms to collaboratively optimize diagnostic model parameters and fusion weights to improve the ability to identify weak early corrosion signals. Ultimately, it achieves accurate diagnosis, location and early warning of corrosion faults, forming an automated intelligent operation and maintenance closed loop.

[0060] The specific embodiments of the present invention will be described below with reference to the accompanying drawings and examples.

[0061] Example 1

[0062] Please see Figure 1 The technical solution for a grounding grid corrosion fault diagnosis method provided in this embodiment includes the following steps:

[0063] S1: By deploying various types of sensors at key points of the grounding grid, the electrical signals and environmental parameters of the grounding grid are collected in real time. The collected raw data is then processed for time synchronization, filtering, noise reduction, and standardization to generate a high-quality multi-source dataset.

[0064] S2: Perform feature extraction on the processed multi-source data, including time-domain and frequency-domain analysis of electrical signals to extract electrical features, statistical calculation of environmental parameters to extract environmental features, and normalize and weighted fuse the extracted multi-source features to generate a comprehensive fault feature vector;

[0065] S3: Construct a diagnostic model suitable for multiple feature inputs, use the Whale Optimization Algorithm (WOA) to adaptively optimize the parameters and feature fusion weights of the diagnostic model, and use the fused comprehensive fault feature vector to train the diagnostic model to obtain the optimized diagnostic model.

[0066] S4: Input the comprehensive fault feature vector into the optimized diagnostic model, output the corrosion fault diagnosis result, and generate a fault report and alarm information based on the diagnosis result.

[0067] In step S1, we selected the grounding grid of a 220kV substation as the monitoring target and deployed monitoring terminals at 15 key nodes of the grounding grid (including the grounding down conductors of the main transformer, surge arresters, circuit breakers, and the edge and central areas of the grounding grid). Each terminal integrates multiple sensors, including one or more of grounding current sensors, voltage sensors, temperature sensors, and humidity sensors.

[0068] In this embodiment, the electrical parameter sensor uses a high-sensitivity zero-flux DC transformer with an accuracy of ±2.5mA to monitor the weak DC component current in the grounding down conductor; at the same time, a 24-bit high-precision ADC acquisition card is used to measure the ground potential difference, and the sampling frequency is set to 10kHz.

[0069] The environmental parameter sensors use digital temperature and humidity sensors (such as SHT35) to measure the temperature and humidity of the surface air; at the same time, soil temperature, humidity and pH sensors are buried near key nodes to obtain more comprehensive environmental information.

[0070] All monitoring terminals are connected to the data acquisition server via on-site industrial Ethernet. Data synchronization is achieved through a GPS timing module, ensuring that the timestamp error of the entire network is less than 1 millisecond.

[0071] After receiving the raw data, the acquisition server performs a preprocessing process. It uses a combination of hardware clock synchronization and software alignment to achieve time synchronization. Each data acquisition node is equipped with a high-precision clock chip to provide a unified time reference for the entire monitoring system.

[0072] The filtering and denoising are performed using a multi-stage filtering strategy, wherein,

[0073] For electrical signals, a moving average filter is used to remove obvious gross errors and impulse noise. Then, a threshold denoising method based on wavelet transform is adopted. An appropriate wavelet basis is selected for multi-scale decomposition. The high-frequency detail coefficients after decomposition are subjected to soft thresholding to effectively suppress white noise and preserve the abrupt change characteristics of the signal.

[0074] For environmental parameter data, a low-pass filter is used to smooth the data, remove high-frequency measurement noise, and retain the true trend of environmental changes.

[0075] The standardization process is performed using Z-score standardization to transform the original data into a standard normal distribution.

[0076] In this embodiment, the electrical signals include current and voltage signals. First, a 50Hz notch filter is used to remove power frequency interference. Then, a 5-layer wavelet packet transform based on the Symlet 8 wavelet basis is applied for decomposition. The high-frequency coefficients of each layer after decomposition are quantized using an improved threshold function (such as the Garrote threshold). Finally, the signal is reconstructed, effectively removing white noise and preserving fault transient characteristics. For slowly varying environmental signals such as temperature and humidity, a first-order low-pass Butterworth filter is used with a cutoff frequency set to 0.1Hz to smooth high-frequency sampling fluctuations.

[0077] Next, sliding window Z-score standardization is used. For a window of length N (e.g., N=3600, corresponding to 1 hour of data), the mean of the data within the window is calculated. ) and standard deviation ( For each data point x within the window, standardize it as follows: This method can adapt to slow time-varying data distributions and is superior to global standardization.

[0078] Then, feature extraction in step S2 is performed, including:

[0079] S21: The electrical features of the electrical signal, including transient amplitude features and spectral energy features, are extracted by wavelet transform or adaptive filtering. At the same time, the mean and variance of the environmental parameters, such as temperature and humidity, are calculated to obtain the environmental features.

[0080] S22: Calculate the cross-correlation coefficient between electrical signals and environmental parameters. This is used to reflect the correlation between electrical signals and environmental parameters, and the calculation formula is expressed as:

[0081] ;

[0082] in, and These are sampled values ​​of electrical signals and environmental parameters, respectively. and These are their average values.

[0083] In this embodiment, we extract the root mean square value, waveform factor, and peak index in the time domain from the denoised electrical signal, and the proportion of 1st to 5th harmonic energy and total spectral entropy in the frequency domain obtained by FFT calculation. We also calculate the mean, variance, and trend (linear fitting slope) of temperature and humidity over one hour. The two signals are coupled, and the Pearson correlation coefficient between the effective current value and soil temperature over a 24-hour period is calculated to quantify the impact of the environment on electrical characteristics.

[0084] Next, the extracted multi-source features are normalized and weighted and fused, including:

[0085] Electrical and environmental characteristics are normalized separately to unify their dimensions. Then, the characteristics are linearly weighted and fused according to their importance using linear weighting or principal component analysis to generate a comprehensive fault feature vector. The formula for the linear weighted fusion is:

[0086] ;

[0087] in, For electrical signal feature vectors, For environmental parameter feature vectors, To merge weights, and The Whale Optimization Algorithm (WOA) is adaptively determined.

[0088] In this embodiment, Min-Max normalization is first used to map all the above features to the [0,1] interval, and then an initial feature vector is constructed. fusion weight Determining this is crucial to this embodiment. As an optimization dimension of the Whale Optimization Algorithm (WOA), it is optimized along with the model parameters. Its fitness function is set to diagnose the model's F1 score on the validation set. Through WOA iteration, an optimal set of weights is eventually found, for example, electrical feature weights. Environmental feature weights This indicates that electrical signals contribute more to corrosion diagnosis in this scenario, and the fusion yields a comprehensive feature vector. .

[0089] Next, step S3 involves constructing the diagnostic model and optimizing its WOA (Warning-Oriented Analysis). The diagnostic model is a multilayer feedforward neural network or a support vector machine. The adaptive optimization of the diagnostic model specifically includes:

[0090] The Whale Optimization Algorithm (WOA) optimizes the parameters and / or feature fusion weights of the diagnostic model through iterative optimization. In each iteration, it optimizes the parameters based on the globally optimal parameter vector of the current iteration. Update the entire parameter vector population, where each parameter vector... The update formula is:

[0091] ;

[0092] in, This represents the magnitude of the difference between the current parameter vector and the globally optimal parameter vector. and To control the coefficients of the parameter vector update direction and step size, , , and A random number in the range [0,1]. It is the convergence factor, which decreases linearly with the number of iterations.

[0093] This embodiment constructs a two-layer bidirectional long short-term memory network (Bi-LSTM) as a diagnostic model to better capture forward and backward dependencies in time series. The model's output layer uses the sigmoid function to output a corrosion failure probability between 0 and 1.

[0094] Through co-optimization using the WOA algorithm, the number of hidden layer units, initial learning rate, L2 regularization coefficient, and feature fusion weights of the Bi-LSTM model are adjusted. Together, they constitute the optimization parameter vector of WOA. The WOA population size is set to 30, and the maximum number of iterations is 100. Convergence factor. The parameter vector decreases linearly from 2 to 0. In each iteration, for each parameter vector (individual whale), a Bi-LSTM model is built and trained with its corresponding parameter configuration, using the validation set F1 score as the fitness value. The WOA updates the global optimum and individual position based on the fitness value, continuously evolving. The final output is the parameter combination that yields the highest F1 score.

[0095] The diagnostic model is trained using a supervised learning mechanism, specifically including:

[0096] The fused integrated fault feature vector As input, the corresponding corrosion fault labels are used as the expected output to construct a training sample set. The cross-entropy loss function is used as the optimization objective for model training. The weighted cross-entropy effectively alleviates the impact of the imbalance between the number of normal and fault state samples in the grounding grid corrosion data on model training by assigning higher weights to fault categories with fewer samples.

[0097] During training, an adaptive moment estimation optimizer is used in conjunction with a cosine annealing strategy to dynamically adjust the learning rate in order to accelerate model convergence and avoid getting trapped in local optima. At the same time, an L2 regularization term is introduced into the loss function of the diagnostic model to constrain the model weights and prevent overfitting. An early stopping strategy is used to automatically terminate training when the performance on the validation set no longer improves.

[0098] During training, the generalization performance of the diagnostic model was evaluated using the K-fold cross-validation method, and model selection was based on a comprehensive index of F1 score and accuracy on the validation set.

[0099] After training is complete, the final diagnostic model parameters, feature fusion weights, and mean and standard deviation parameters required for data preprocessing are saved to form a complete and deployable diagnostic model.

[0100] In this embodiment, the final model is retrained on the training set using the Adam optimizer and FocalLoss loss function with optimized parameter configuration to address the problem of scarce corrosion samples.

[0101] After processing the real-time data stream through the above process, it is input into the trained optimized model to output diagnostic results. First, the corrosion index is defined. The degree of corrosion is quantified as follows:

[0102] ;

[0103] in, This refers to the grounding resistance under normal conditions. For the current measurement of grounding resistance;

[0104] Then based on the corrosion index The system combines the model output results to determine whether a corrosion fault has occurred and the degree of corrosion. If the diagnostic indicators exceed the preset threshold, it is determined to be a corrosion fault and the fault location is located.

[0105] In addition, the generation of fault reports and alarm information includes:

[0106] It automatically generates diagnostic reports, uploads fault type, location, and severity information to the monitoring platform via the communication module, triggers on-site alarm devices, and provides a visual interface to display diagnostic results and historical trends in real time, supporting maintenance personnel in developing repair plans.

[0107] In this embodiment, if the output corrosion failure probability is greater than 0.85 (adjustable threshold), it is determined to be a failure. Simultaneously, the corrosion index is calculated. The system automatically locates the three monitoring nodes with the highest probability and highlights them in red on the monitoring screen. At the same time, it generates an alarm work order and pushes it to the maintenance personnel via SMS and system message.

[0108] To verify the effectiveness of the method of this invention, it was compared with the method of this invention, a fixed-parameter Bi-LSTM model without WOA optimization, and the traditional support vector machine (SVM) method. In a year-long field data test, the method of this invention achieved a fault detection rate of 98.5%, a false negative rate of 1.2%, and a false positive rate of only 2.1%, all significantly better than the comparative methods, demonstrating its superior diagnostic accuracy and reliability in complex field environments.

[0109] Based on this, this embodiment also provides a grounding grid corrosion fault diagnosis system, such as... Figure 2 As shown, it includes:

[0110] The data acquisition and preprocessing module 10 is used to acquire electrical signals and environmental parameter data of the grounding grid in real time by deploying various types of sensors at key points of the grounding grid, and to perform time synchronization, filtering and noise reduction and standardization processing on the acquired raw data to generate high-quality multi-source datasets.

[0111] The multi-source feature extraction and fusion module 20 is used to extract features from the processed multi-source data, including performing time-domain and frequency-domain analysis on electrical signals to extract electrical features, performing statistical calculations on environmental parameters to extract environmental features, and normalizing and weighting the extracted multi-source features to generate a comprehensive fault feature vector.

[0112] The diagnostic model construction and optimization module 30 is used to construct a diagnostic model suitable for multiple feature inputs. The Whale Optimization Algorithm (WOA) is used to adaptively optimize the parameters and feature fusion weights of the diagnostic model. The fused comprehensive fault feature vector is used to train the diagnostic model to obtain the optimized diagnostic model.

[0113] The fault diagnosis and result output module 40 is used to input the comprehensive fault feature vector into the optimized diagnostic model, output the corrosion fault diagnosis result, and generate a fault report and alarm information based on the diagnosis result.

[0114] It should be noted that the steps in the grounding grid corrosion fault diagnosis method provided in this embodiment can be implemented based on the corresponding modules in the grounding grid corrosion fault diagnosis system. Those skilled in the art can refer to the technical solution of the system to implement the steps of the method. That is, the embodiments in the system can be understood as preferred examples of implementing the method, and will not be elaborated here.

[0115] This embodiment also provides an electronic device, such as... Figure 3 As shown, the electronic device includes a processor 14 and a memory 13. The memory 13 stores machine-executable instructions that can be executed by the processor 14, which executes the machine-executable instructions to implement the above-described audio control method.

[0116] Furthermore, Figure 3 The electronic device shown also includes a bus 12 and a communication interface 11, with the processor 14, the communication interface 11 and the memory 13 connected via the bus 12.

[0117] The memory 13 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 11 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 12 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0118] Processor 14 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 14 or by instructions in software form. Processor 14 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in this embodiment. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this embodiment can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 13. The processor 14 reads the information in memory 13 and, in conjunction with its hardware, completes the steps of the grounding grid corrosion fault diagnosis method.

[0119] This disclosure also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, storing a computer program that, when run on a computer, causes the computer to perform the steps of a grounding grid corrosion fault diagnosis method.

[0120] Finally, it should be noted that the above description is only a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be pointed out that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

[0121] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for diagnosing corrosion faults in grounding grids, characterized in that, Includes the following steps: S1: By deploying various types of sensors at key points of the grounding grid, the electrical signals and environmental parameters of the grounding grid are collected in real time. The collected raw data is then processed for time synchronization, filtering, noise reduction, and standardization to generate a high-quality multi-source dataset. S2: Perform feature extraction on the processed multi-source data, including time-domain and frequency-domain analysis of electrical signals to extract electrical features, statistical calculation of environmental parameters to extract environmental features, and normalize and weighted fuse the extracted multi-source features to generate a comprehensive fault feature vector; The process of normalizing and weighted fusion of the extracted multi-source features includes: Electrical and environmental characteristics are normalized separately to unify their dimensions. Then, the characteristics are linearly weighted and fused according to their importance using linear weighting or principal component analysis to generate a comprehensive fault feature vector. The formula for the linear weighted fusion is: ； in, For electrical signal feature vectors, For environmental parameter feature vectors, To merge weights, and Determined adaptively using the Whale Optimization Algorithm (WOA); S3: Construct a diagnostic model suitable for multiple feature inputs, use the Whale Optimization Algorithm (WOA) to adaptively optimize the parameters and feature fusion weights of the diagnostic model, and use the fused comprehensive fault feature vector to train the diagnostic model to obtain the optimized diagnostic model. The diagnostic model is a multilayer feedforward neural network or a support vector machine, and the adaptive optimization of the diagnostic model specifically includes: The Whale Optimization Algorithm (WOA) optimizes the parameters and / or feature fusion weights of the diagnostic model through iterative optimization. In each iteration, it optimizes the parameters based on the globally optimal parameter vector of the current iteration. Update the entire parameter vector population, where each parameter vector... The update formula is: ； in, This represents the magnitude of the difference between the current parameter vector and the globally optimal parameter vector. and To control the coefficients of the parameter vector update direction and step size, , , and A random number in the range [0,1]. The convergence factor decreases linearly with the number of iterations. S4: Input the comprehensive fault feature vector into the optimized diagnostic model, output the corrosion fault diagnosis result, and generate a fault report and alarm information based on the diagnosis result.

2. The grounding grid corrosion fault diagnosis method according to claim 1, characterized in that, In step S2, the feature extraction specifically includes: S21: The electrical features of the electrical signal, including transient amplitude features and spectral energy features, are extracted by wavelet transform or adaptive filtering. At the same time, the mean and variance of the environmental parameters, such as temperature and humidity, are calculated to obtain the environmental features. S22: Calculate the cross-correlation coefficient between electrical signals and environmental parameters. This is used to reflect the correlation between electrical signals and environmental parameters, and the calculation formula is expressed as: ； in, and These are sampled values ​​of electrical signals and environmental parameters, respectively. and These are their average values.

3. The grounding grid corrosion fault diagnosis method according to claim 1, characterized in that, In step S3, the diagnostic model is trained using a supervised learning mechanism, specifically including: The fused integrated fault feature vector As input, the corresponding corrosion fault labels are used as the expected output to construct a training sample set, and the cross-entropy loss function is used as the optimization objective for model training. During training, an adaptive moment estimation optimizer is used in conjunction with a cosine annealing strategy to dynamically adjust the learning rate in order to accelerate model convergence and avoid getting trapped in local optima. At the same time, an L2 regularization term is introduced into the loss function of the diagnostic model to constrain the model weights and prevent overfitting. An early stopping strategy is used to automatically terminate training when the performance on the validation set no longer improves. During training, the generalization performance of the diagnostic model was evaluated using the K-fold cross-validation method, and model selection was based on a comprehensive index of F1 score and accuracy on the validation set. After training is complete, the final diagnostic model parameters, feature fusion weights, and mean and standard deviation parameters required for data preprocessing are saved to form a complete and deployable diagnostic model.

4. The grounding grid corrosion fault diagnosis method according to claim 1, characterized in that, In step S4, the output corrosion fault diagnosis results specifically include: Define corrosion index The degree of corrosion is quantified as follows: ； in, This refers to the grounding resistance under normal conditions. For the current measurement of grounding resistance; According to the corrosion index The system combines the model output results to determine whether a corrosion fault has occurred and the degree of corrosion. If the diagnostic indicators exceed the preset threshold, it is determined to be a corrosion fault and the fault location is located.

5. The grounding grid corrosion fault diagnosis method according to claim 1, characterized in that, In step S1, the various types of sensors include one or more of ground current sensors, voltage sensors, temperature sensors, and humidity sensors; The time synchronization is achieved by combining hardware clock synchronization and software alignment. Each data acquisition node is equipped with a high-precision clock chip to provide a unified time reference for the entire monitoring system. The filtering and denoising are performed using a multi-stage filtering strategy, wherein, For electrical signals, a moving average filter is used to remove obvious gross errors and impulse noise. Then, a threshold denoising method based on wavelet transform is adopted. An appropriate wavelet basis is selected for multi-scale decomposition. The high-frequency detail coefficients after decomposition are subjected to soft thresholding to effectively suppress white noise and preserve the abrupt change characteristics of the signal. For environmental parameter data, a low-pass filter is used to smooth the data, remove high-frequency measurement noise, and retain the true trend of environmental changes. The standardization process is performed using Z-score standardization to transform the original data into a standard normal distribution.

6. The grounding grid corrosion fault diagnosis method according to claim 1, characterized in that, In step S4, generating fault reports and alarm information includes: It automatically generates diagnostic reports, uploads fault type, location, and severity information to the monitoring platform via the communication module, triggers on-site alarm devices, and provides a visual interface to display diagnostic results and historical trends in real time, supporting maintenance personnel in developing repair plans.

7. A grounding grid corrosion fault diagnosis system, characterized in that, include: The data acquisition and preprocessing module is used to collect electrical signals and environmental parameter data of the grounding grid in real time by deploying various types of sensors at key points of the grounding grid, and to perform time synchronization, filtering and noise reduction and standardization processing on the collected raw data to generate high-quality multi-source datasets. The multi-source feature extraction and fusion module is used to extract features from the processed multi-source data, including time-domain and frequency-domain analysis of electrical signals to extract electrical features, statistical calculation of environmental parameters to extract environmental features, and normalization and weighted fusion of the extracted multi-source features to generate a comprehensive fault feature vector. The process of normalizing and weighted fusion of the extracted multi-source features includes: Electrical and environmental characteristics are normalized separately to unify their dimensions. Then, the characteristics are linearly weighted and fused according to their importance using linear weighting or principal component analysis to generate a comprehensive fault feature vector. The formula for the linear weighted fusion is: ； in, For electrical signal feature vectors, For environmental parameter feature vectors, To merge weights, and Determined adaptively using the Whale Optimization Algorithm (WOA); The diagnostic model construction and optimization module is used to construct a diagnostic model suitable for multiple feature inputs. The Whale Optimization Algorithm (WOA) is used to adaptively optimize the parameters and feature fusion weights of the diagnostic model. The fused comprehensive fault feature vector is used to train the diagnostic model to obtain the optimized diagnostic model. The diagnostic model is a multilayer feedforward neural network or a support vector machine, and the adaptive optimization of the diagnostic model specifically includes: The Whale Optimization Algorithm (WOA) optimizes the parameters and / or feature fusion weights of the diagnostic model through iterative optimization. In each iteration, it optimizes the parameters based on the globally optimal parameter vector of the current iteration. Update the entire parameter vector population, where each parameter vector... The update formula is: ； in, This represents the magnitude of the difference between the current parameter vector and the globally optimal parameter vector. and To control the coefficients of the parameter vector update direction and step size, , , and A random number in the range [0,1]. The convergence factor decreases linearly with the number of iterations. The fault diagnosis and result output module is used to input the comprehensive fault feature vector into the optimized diagnostic model, output the corrosion fault diagnosis result, and generate a fault report and alarm information based on the diagnosis result.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the grounding grid corrosion fault diagnosis method as described in any one of claims 1-6.

9. An electronic device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the grounding grid corrosion fault diagnosis method as described in any one of claims 1-6.

Images