Single-phase earth fault rapid positioning method based on multi-source data fusion

Abstract

The invention relates to the technical field of power failure troubleshooting, and discloses a single-phase earth fault rapid positioning method based on multi-source data fusion, and the method comprises the steps: collecting historical fault data, power system operation data and environment data through employing a plurality of monitoring devices, such as a power system monitoring device, a fault recorder and a distributed sensor; and forming a basic data set. By performing fault feature analysis on the data set, a dual machine learning model is constructed, including a fault prediction model and a positioning model. A training data set is used to train a model, model parameters are optimized, then the trained model is used to predict real-time data, potential fault risks are identified, and a targeted troubleshooting plan and a rapid positioning strategy are formulated. According to the invention, through data fusion and a machine learning algorithm, the accuracy of fault positioning is improved, the positioning time is shortened, and the power failure time and the influence on users are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power fault troubleshooting, and in particular to a method for quickly locating a single-phase grounding fault based on multi-source data fusion. Background Art

[0002] In the power system, single-phase grounding fault is a common fault type. Its rapid and accurate positioning is crucial to ensure the stable operation of the power system and timely restoration of power supply.

[0003] At present, the single-phase grounding fault location method mainly relies on manual inspection and simple fault indicators. Manual inspection is time-consuming and laborious, and is greatly affected by external factors such as weather and terrain, and it is often difficult to find the fault point in a timely and accurate manner. Although the fault indicator can provide certain fault information, its accuracy and reliability are limited, especially in the complex and changeable power system environment, and its positioning effect is often unsatisfactory. For example, in a single-phase grounding fault that occurred in an actual power system, because the fault point was located in a remote mountainous area, it was difficult for manual inspection to arrive in time, resulting in a fault location time of up to several hours, which seriously affected the normal operation of the power system and the user's electricity experience. At the same time, due to the false alarm and missed alarm of the fault indicator, the accuracy of fault location is greatly reduced, further extending the time for fault handling.

[0004] The existing single-phase grounding fault location method still lacks the effective fusion and utilization of data from different sources. There are many types of monitoring equipment in the power system, including power system monitoring devices, fault recorders and distributed sensors. The data collected by these devices covers all aspects of the power system. By fusing and analyzing these data, the accuracy of fault location and the early warning capability of faults can be improved. Summary of the invention

[0005] The purpose of the present invention is to provide a method for quickly locating a single-phase grounding fault based on multi-source data fusion to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for quickly locating a single-phase grounding fault based on multi-source data fusion, the method comprising:

[0007] Use a variety of monitoring equipment, including power system monitoring devices, fault recorders and distributed sensors to collect historical single-phase grounding fault data, as well as related power system operation data and environmental data, and pre-process the collected data to form a single-phase grounding fault basic data set;

[0008] Perform fault feature analysis on the basic data set, including fault occurrence period analysis, fault type pattern recognition and system operation status impact analysis, and merge the analysis result data with the basic data set to form a single-phase grounding fault training data set;

[0009] The design of dual machine learning models is adopted to construct a fault prediction model and a positioning model respectively, and a machine learning algorithm is used to fit the occurrence, location and fault degree of single-phase grounding faults, and the outputs of the two models are used to build a comprehensive decision-making model; the fault prediction model is used to predict the probability and possible time period of single-phase grounding faults in the power system, and the positioning model is used to determine the specific location of the fault;

[0010] The dual machine learning model is trained using the single-phase ground fault training dataset, and the model parameters are optimized through cross-validation and gradient descent algorithms;

[0011] Use the trained model to predict real-time power system data, identify potential single-phase grounding fault risks, and develop targeted fault troubleshooting plans and rapid location strategies based on the prediction results;

[0012] Apply the developed troubleshooting plan and rapid location strategy to the actual power system, and continuously fine-tune and optimize the strategy through a real-time monitoring system and feedback adjustment mechanism.

[0013] Preferably, the method for implementing the fault occurrence period analysis includes:

[0014] Extract the timestamp information from the preprocessed single-phase grounding fault basic data set, sort and segment it according to the time dimension to form time series data;

[0015] The autoregressive integrated moving average model (ARIMA) is used to model the time series data and capture the temporal regularity and periodic characteristics of fault occurrence.

[0016] Preferably, the data preprocessing method includes data cleaning, which specifically includes:

[0017] The missing values ​​in the original data set were processed and the mean imputation method was used to estimate and supplement the missing values;

[0018] Detect and process outliers in the original data set, use the threshold judgment method to identify outliers, and choose to delete outliers, replace them with statistical values, or perform outlier smoothing according to the specific situation;

[0019] Identify and remove duplicate data in the original dataset.

[0020] Preferably, the data preprocessing method also includes:

[0021] Standardize the continuous data to eliminate the dimension difference. Use the Z-score method to standardize the data. Subtract the mean of each numerical data and divide it by its standard deviation so that the processed data conforms to the standard normal distribution. The formula for standardization is: Z = (X-μ) / σ; where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data;

[0022] Convert the format and type of the original data set, including converting strings into numeric data, unifying date and time data in different formats into a standard date and time format, and performing one-hot encoding conversion on categorical data.

[0023] Preferably, the method for implementing fault type pattern recognition includes:

[0024] The fault data in the preprocessed single-phase grounding fault basic data set is subjected to feature extraction, and the extracted features include the magnitude of the fault current, the rate of change of the fault voltage, the fault duration, and the system state parameters before and after the fault occurs;

[0025] The K-means clustering algorithm is used to perform cluster analysis on the extracted features to identify different fault type patterns; the goal of the clustering algorithm is to minimize the intra-class differences and maximize the inter-class differences, so as to accurately classify different fault types.

[0026] Preferably, the method for implementing the system operation status impact analysis includes:

[0027] Extracting features from the system operating status data in the preprocessed single-phase grounding fault basic data set, wherein the features include system voltage, current, frequency, load rate, reactive power factor and operating status parameters of key equipment;

[0028] The Pearson correlation coefficient analysis algorithm is used to calculate the correlation between the extracted system operation status characteristics and the occurrence of single-phase grounding faults. By quantifying the degree of correlation between the characteristics and the faults, the system operation status factors that have a significant impact on the occurrence of the faults are identified. The calculation formula of the Pearson correlation coefficient r is:

[0029]

[0030] Among them, r represents the correlation coefficient between the feature and the fault, n represents the number of data points, x represents the feature value, and n represents the probability value of the fault occurrence.

[0031] Preferably, the steps of building the dual machine learning model include:

[0032] A fault prediction model is constructed for the prediction of the occurrence of single-phase grounding faults. The model adopts a logistic regression classification algorithm, with the input being the feature vector of power system operation data, environmental data and historical fault data, and the output being the occurrence probability and possible time period of single-phase grounding faults. The formula of the logistic regression model is:

[0033]

[0034] Where P(Y=1|X) represents the probability of a single-phase ground fault occurring under a given input feature vector X; β 0 ,β 1 ,…,β n are model parameters, learned through training data;

[0035] To determine the location of the single-phase grounding fault, a positioning model is constructed; the model uses the support vector regression SVR algorithm, the input is the power system state data when the fault occurs, the data recorded by the fault recorder and the data of the distributed sensor, and the output is the specific location of the fault; the formula of the SVR model is:

[0036]

[0037] Where f(x) represents the predicted value of the fault location; α i and is the Lagrange multiplier, which is learned through training data; K(x i ,x) is the kernel function, which is used to measure the similarity between input feature vectors; b is the bias term.

[0038] Preferably, in the process of building a dual machine learning model, cross-validation and gradient descent algorithms are used to optimize model parameters; cross-validation is used to evaluate the generalization ability of the model to prevent overfitting; the gradient descent algorithm is used to iteratively update the model parameters so that the loss function of the model on the training data reaches the minimum value.

[0039] Preferably, a targeted fault troubleshooting plan is formulated based on the output results of the fault prediction model; if a high risk of single-phase grounding fault is predicted within a certain period of time, the monitoring and inspection of the power system is strengthened within that period of time.

[0040] Preferably, a rapid positioning strategy is formulated based on the output results of the positioning model; once a single-phase grounding fault actually occurs, the real-time data at the time of the fault is immediately input into the positioning model to quickly and accurately determine the fault location.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The present invention forms a comprehensive single-phase ground fault basic data set by integrating data collected by various monitoring devices such as power system monitoring devices, fault recorders and distributed sensors. By analyzing the fault characteristics of these data and building a dual machine learning model for training and prediction, it is possible to more accurately identify fault characteristics, thereby improving the accuracy of fault location. Compared with the traditional method that relies on manual inspections and simple fault indicators, the present invention has higher positioning accuracy and significantly reduces false alarm and missed alarm rates.

[0043] The present invention uses machine learning algorithms to predict real-time power system data, which can promptly detect potential single-phase grounding fault risks and quickly determine the specific location of the fault, greatly shortening the time for fault locating, allowing maintenance personnel to quickly reach the fault point for processing, thereby reducing the power outage time of the power system and the impact on users. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a processing step diagram of a method for quickly locating a single-phase grounding fault based on multi-source data fusion according to the present invention;

[0045] Figure 2 A schematic diagram of the design of a dual machine learning model;

[0046] Figure 3 This is a step diagram for analyzing the impact of system operation status. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] See also Figure 1-3 The present invention provides a technical solution: a method for quickly locating a single-phase grounding fault based on multi-source data fusion, the method comprising:

[0049] A variety of monitoring equipment, including power system monitoring devices, fault recorders and distributed sensors, is used to collect all-round data on the power system. The monitoring equipment is used to collect historical single-phase grounding fault data in real time or regularly, as well as related power system operation data and environmental data. The collected data is preprocessed to form a single-phase grounding fault basic data set.

[0050] The data preprocessing methods include:

[0051] i) Process the missing values ​​in the original data set and use the mean imputation method to estimate and supplement the missing values.

[0052] ii) Detect and process outliers in the original data set, use the threshold judgment method to identify outliers, and choose to delete outliers, replace them with statistical values, or perform outlier smoothing according to the specific situation;

[0053] iii) Identify and remove duplicate data in the original dataset.

[0054] iv) Standardize the continuous data to eliminate dimensional differences. Use the Z-score method to standardize the data. Subtract the mean of each numerical data and divide it by its standard deviation so that the processed data conforms to the standard normal distribution. The standardization formula is: Z = (X-μ) / σ; where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data.

[0055] v) Convert the format and type of the original data set, including converting strings into numeric data, unifying date and time data in different formats into a standard date and time format; and performing one-hot encoding conversion on categorical data.

[0056] vi) Conduct in-depth fault feature analysis on the basic data set, including fault occurrence period analysis to identify the time pattern and trend of fault occurrence; fault type pattern recognition to distinguish different types of single-phase grounding faults; and system operation status impact analysis to evaluate the impact of power system operation status on fault occurrence. The analysis result data is merged with the basic data set to form a more comprehensive single-phase grounding fault training data set.

[0057] The design of dual machine learning models is adopted to build fault prediction model and location model respectively. The fault prediction model uses machine learning algorithm to predict the probability and possible time period of single-phase grounding fault to identify potential fault risks in advance. The location model is used to determine the specific location of the fault and accurately locate the fault point by analyzing the data of different monitoring devices. The two models work together to build a comprehensive decision-making model to improve the accuracy and efficiency of fault location.

[0058] After the model is built, the dual machine learning model is trained using the single-phase ground fault training dataset. The model parameters are optimized through cross-validation and gradient descent algorithms to ensure the performance and stability of the model.

[0059] Apply the trained model to the actual power system to predict and analyze the real-time power system data. According to the prediction results, formulate targeted fault troubleshooting plans and rapid location strategies. Through the real-time monitoring system and feedback adjustment mechanism, continuously fine-tune and optimize the strategy to adapt to the changes in the power system operation status and the diversity of fault types.

[0060] This outbreak is further described below in conjunction with Examples 1 to 3:

[0061] Embodiment 1:

[0062] This embodiment is used to describe the implementation method of performing fault feature analysis on a basic data set, specifically:

[0063] The implementation methods of fault occurrence period analysis include:

[0064] The timestamp information in the preprocessed single-phase ground fault basic data set is extracted, sorted and segmented according to the time dimension to form time series data; according to the analysis requirements, the sorted timestamp information is divided into different time periods. These time periods can be fixed (such as daily, weekly, monthly, etc.) or dynamic (such as divided according to the frequency or mode of fault occurrence).

[0065] The original data set is divided into multiple time series data segments according to the time segmentation results. Each data segment contains the parameter information and timestamp of all faults in the time period.

[0066] The autoregressive integrated moving average model (ARIMA) is selected as the model for time series analysis. The ARIMA model can capture trends, seasonality, and random fluctuations in time series data, and is suitable for analyzing the temporal regularity and periodic characteristics of fault occurrence.

[0067] According to the characteristics of time series data, determine the parameters of the ARIMA model, including the order of the autoregressive term (AR), the order of the moving average term (MA), and the number of differences (d). Use the segmented time series data to train the ARIMA model. By fitting the model, the temporal regularity and periodicity characteristics in the time series data can be captured. During the training process, the model parameters can be adjusted to minimize the error of the model on the training data.

[0068] The implementation methods of fault type pattern recognition include:

[0069] The basic data set of single-phase grounding fault is collected from the power system. The data set contains electrical parameter information such as current and voltage when the fault occurs, as well as system state parameters before and after the fault, such as load conditions, generator output power, etc. According to the characteristics of single-phase grounding fault, the fault current magnitude, fault voltage change rate, fault duration and system state parameters before and after the fault are selected as features for extraction. For each fault data point, its corresponding characteristic value is calculated. For example, the fault current magnitude can be obtained by measuring the current peak value or effective value when the fault occurs; the fault voltage change rate can be obtained by calculating the ratio of the voltage difference before and after the fault to the voltage before the fault; the fault duration can be obtained by recording the time interval from the occurrence to the end of the fault; the system state parameters before and after the fault are selected and calculated according to the actual situation.

[0070] The extracted feature values ​​are arranged in the order of the fault data points to form a feature data set. The K-means clustering algorithm is selected to perform cluster analysis on the feature data set. The K-means algorithm is a commonly used clustering algorithm, whose goal is to minimize the intra-class differences and maximize the inter-class differences, so as to accurately classify different fault types.

[0071] According to the characteristics of the feature data set and the analysis requirements, determine the clustering parameters of the K-means algorithm, including the number of clusters K and the method for selecting the initial cluster center. The selection of K can be determined by trying different K values ​​and comparing the clustering results; the method for selecting the initial cluster center can be random selection or based on some heuristic strategy.

[0072] Clustering process: Use the K-means algorithm to perform cluster analysis on the feature data set. The algorithm will gradually converge to a stable clustering result by iteratively updating the cluster center and redistributing data points based on the selected K value and initial cluster center. During the iteration process, the changes in the cluster center and the distribution of data points should be monitored to ensure the stability and accuracy of the clustering process.

[0073] The fault type pattern obtained by cluster analysis is applied to the actual power system. In the process of fault prediction and location, the clustering results are used to identify the type pattern of the current fault.

[0074] The implementation methods of system operation status impact analysis include:

[0075] The system operation status data in the preprocessed single-phase ground fault basic data set is feature extracted, and the features include system voltage, current, frequency, load rate, reactive power factor and operation status parameters of key equipment; for each data point, the corresponding system operation status characteristic value is calculated. For example, the system voltage can be obtained by measuring the bus voltage; the current can be obtained by measuring the line current; the frequency can be obtained by measuring the system frequency; the load rate can be obtained by calculating the ratio of the actual load to the rated load; the reactive power factor can be obtained by measurement or calculation; the operation status parameters of key equipment are selected and calculated according to the actual situation of the equipment.

[0076] The Pearson correlation coefficient analysis algorithm is selected to calculate the correlation between the extracted system operation status characteristics and the occurrence of single-phase grounding faults. The Pearson correlation coefficient can quantify the linear relationship between two variables, and its value range is between -1 and 1, indicating a complete negative correlation to a complete positive correlation. According to the calculation formula of the Pearson correlation coefficient, the correlation coefficient between each system operation status characteristic and the occurrence of a single-phase grounding fault is calculated. During the calculation process, it is necessary to traverse all data points in the data set, and calculate the sum of the products between the characteristic values ​​and the fault occurrence probability values, the sum of the squares of the characteristic values, and the sum of the squares of the fault occurrence probability values, and finally obtain the correlation coefficient. The calculation formula for the Pearson correlation coefficient r is:

[0077]

[0078] Among them, r represents the correlation coefficient between the feature and the fault, n represents the number of data points, x represents the feature value, and n represents the probability value of the fault occurrence.

[0079] According to the calculated correlation coefficient value, the correlation between the system operation status characteristics and the occurrence of single-phase grounding fault is determined. Generally, the larger the absolute value of the correlation coefficient, the stronger the correlation between the characteristics and the fault; a positive correlation coefficient indicates a positive correlation between the characteristics and the fault; a negative correlation coefficient indicates a negative correlation between the characteristics and the fault.

[0080] The calculated correlation coefficients are sorted and analyzed to identify the system operation status factors that have a significant impact on the occurrence of single-phase grounding faults.

[0081] Embodiment 2:

[0082] This embodiment is used to describe a method for constructing a dual machine learning model, specifically:

[0083] The steps to build a fault prediction model include:

[0084] Step 1: Collect operation data from the power system, including real-time parameters such as voltage, current, and power, as well as environmental data such as ambient temperature, humidity, and wind speed that may affect the stability of the power system. At the same time, organize historical fault data, especially cases of single-phase grounding faults, including fault type, occurrence time, duration, and other information.

[0085] The collected data are cleaned to remove outliers and missing values. The data are normalized to eliminate dimensional differences and make the data comparable. The feature vectors related to single-phase grounding faults, such as voltage fluctuations and abnormal current increases, are extracted as inputs to the logistic regression model.

[0086] Step 2: Select the logistic regression classification algorithm as the basis of the fault prediction model. Logistic regression is a widely used classification algorithm suitable for binary classification problems and can output the probability of a single-phase ground fault under a given input feature vector. The formula of the logistic regression model is:

[0087]

[0088] Where P(Y=1|X) represents the probability of a single-phase ground fault occurring under a given input feature vector X; β 0 ,β 1 ,…,β n are model parameters, which are learned through training data.

[0089] Step 3: Divide the preprocessed data into a training dataset and a validation dataset. The training dataset is used to train the logistic regression model, and the validation dataset is used to evaluate the performance of the model. Use the training dataset to train the logistic regression model. Use optimization algorithms such as gradient descent or Newton iteration to solve the model parameter β 0 ,β 1 ,…,β n During the training process, the model parameters are continuously adjusted to minimize the loss function of the model on the training data set.

[0090] Step 4: Use the validation data set to evaluate the performance of the logistic regression model. Calculate the model's accuracy, recall, and F1 score indicators to measure the model's ability to predict the occurrence of single-phase grounding faults. Based on the validation results, optimize the logistic regression model. The prediction accuracy of the model can be improved by adjusting model parameters, increasing training samples, and improving feature vector extraction.

[0091] The steps to build the positioning model include:

[0092] Step 1: When a single-phase grounding fault occurs, quickly collect power system status data, including voltage and current changes before and after the fault, and high-frequency transient signals recorded by the fault recorder. At the same time, obtain data from distributed sensors, such as temperature and vibration on the line. Clean the collected data to remove invalid and redundant information. Synchronize the data to ensure that data from different devices remain consistent in time. Extract features related to the fault location, such as the mutation point of the current waveform and the energy distribution of the high-frequency signal, as input to the SVR model.

[0093] Step 2: Select the support vector regression (SVR) algorithm as the basis of the location model. SVR is a regression algorithm based on support vector machines, which is suitable for processing nonlinear regression problems and can output the predicted value of the fault location. The formula of the SVR model is:

[0094]

[0095] Where f(x) represents the predicted value of the fault location; α i and is the Lagrange multiplier, which is learned through training data; K(x i ,x) is the kernel function, which is used to measure the similarity between input feature vectors; b is the bias term.

[0096] Step 3: Divide the preprocessed data into a training data set and a test data set. The training data set is used to train the SVR model, and the test data set is used to evaluate the performance of the model. The SVR model is trained using the training data set. By solving the optimization problem, the Lagrange multiplier α is obtained. i , And bias term b. During the training process, the model parameters are continuously adjusted to minimize the error of the model on the training data set.

[0097] Step 4: Use the test data set to evaluate the performance of the SVR model. Calculate the model's positioning accuracy, mean square error and other indicators to measure the model's ability to predict the location of a single-phase grounding fault.

[0098] Step 5: Apply the trained SVR model to the actual power system. When a single-phase grounding fault occurs, quickly collect relevant data and input it into the model to predict the specific location of the fault.

[0099] Here is a Python code example for building a dual machine learning model:

[0100] import numpy as np

[0101] import pandas as pd

[0102] from sklearn.linear_model import LogisticRegression

[0103] from sklearn.svm import SVR

[0104] from sklearn.model_selection import train_test_split

[0105] from sklearn.metrics import accuracy_score,mean_squared_error

[0106] #electricity_data: Power system operation data

[0107] #environment_data:Environment data

[0108] #historical_fault_data: Historical fault data

[0109] #fault_state_data: Power system status data during fault

[0110] #fault_recorder_data: Fault recorder data

[0111] #distributed_sensor_data: Distributed sensor data

[0112] # Divide into training set and test set

[0113] X_train_prediction,

[0114] X_train_location,

[0115] #Build a fault prediction model (logistic regression)

[0116] fault_prediction_model=LogisticRegression()

[0117] fault_prediction_model.fit(X_train_prediction,y_train_prediction)

[0118] #Predict the probability of failure

[0119] y_pred_probability = fault_prediction_model.predict_proba(X_test_prediction)[:,1]#Get the probability of a fault #Build a positioning model (Support Vector Regression SVR)

[0120] fault_location_model = SVR(kernel = 'rbf') # Use radial basis function kernel

[0121] fault_location_model.fit(X_train_location,y_train_location)

[0122] #Predict fault location

[0123] y_pred_location=fault_location_model.predict(X_test_location)

[0124] #Evaluate model performance

[0125] accuracy = accuracy_score(y_test_prediction, fault_prediction_model.predict(X_test_prediction)) #Accuracy of fault prediction model

[0126] mse = mean_squared_error (y_test_location, y_pred_location) # Mean square error of the positioning model

[0127] Integrate the positioning model with the monitoring system of the power system. Realize real-time data collection, processing and model prediction. Set up an early warning mechanism in the monitoring system. When a single-phase grounding fault is predicted to occur, issue an early warning signal in time and display the predicted fault location.

[0128] Embodiment 3:

[0129] This embodiment provides an application example of the present invention:

[0130] Taking a power system as an example, the real-time data set includes power system operation data such as voltage, current, frequency, environmental data such as temperature and humidity, and line status data from distributed sensors. These data are input into the dual machine learning model. The fault prediction model will output the probability and possible time period of single-phase grounding fault, the positioning model will output the approximate location of the fault, and the data fusion model will integrate these data to improve the accuracy of prediction and positioning.

[0131] According to the output of the fault prediction model, formulate a targeted troubleshooting plan. Assume that the fault prediction model predicts that the probability of single-phase grounding fault in a certain area (such as area A) in the next week is high, reaching more than 80%. Based on this prediction result, formulate a fault troubleshooting plan and give priority to the inspection of the lines and equipment in this area. The inspection content includes checking whether the line is aging, whether the insulation is damaged, whether the equipment is operating normally, etc., and record the inspection results to form a list of potential fault points.

[0132] According to the output of the fault prediction model and the positioning model, combined with the inspection results, a rapid positioning strategy is formulated. Assume that at 3 pm on a certain day, real-time data monitors a single-phase grounding fault in area A. Immediately start the rapid positioning process. First, according to the output of the positioning model, determine that the approximate location of the fault is on a line in area A. Then, combined with the list of potential fault points recorded in the inspection record, prioritize these potential fault points. For example, the inspection record shows that the insulation is damaged somewhere on the line, and immediately arrange personnel to go there for inspection and repair. Through the feedback of the real-time monitoring system, confirm that the fault point has been accurately located and eliminated, and at the same time, adjust and optimize the rapid positioning strategy in real time to ensure that subsequent faults can be located and eliminated in a timely and accurate manner.

[0133] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0134] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A single-phase grounding fault rapid location method based on multi-source data fusion, characterized in that: The method comprises: Use a variety of monitoring equipment, including power system monitoring devices, fault recorders and distributed sensors to collect historical single-phase grounding fault data, as well as related power system operation data and environmental data, and pre-process the collected data to form a single-phase grounding fault basic data set; Perform fault feature analysis on the basic data set, including fault occurrence period analysis, fault type pattern recognition and system operation status impact analysis, and merge the analysis result data with the basic data set to form a single-phase grounding fault training data set; The design of dual machine learning models is adopted to construct a fault prediction model and a positioning model respectively, and a machine learning algorithm is used to fit the occurrence, location and fault degree of single-phase grounding faults, and the outputs of the two models are used to build a comprehensive decision-making model; the fault prediction model is used to predict the probability and possible time period of single-phase grounding faults in the power system, and the positioning model is used to determine the specific location of the fault; The dual machine learning model is trained using the single-phase ground fault training dataset, and the model parameters are optimized through cross-validation and gradient descent algorithms; Use the trained model to predict real-time power system data, identify potential single-phase grounding fault risks, and develop targeted fault troubleshooting plans and rapid location strategies based on the prediction results; Apply the developed troubleshooting plan and rapid location strategy to the actual power system, and continuously fine-tune and optimize the strategy through a real-time monitoring system and feedback adjustment mechanism.

2. A single-phase grounding fault rapid location method based on multi-source data fusion according to claim 1, characterized in that: The implementation methods of fault occurrence period analysis include: Extract the timestamp information from the preprocessed single-phase grounding fault basic data set, sort and segment it according to the time dimension to form time series data; The autoregressive integrated moving average model (ARIMA) is used to model the time series data and capture the temporal regularity and periodic characteristics of fault occurrence.

3. The method for quickly locating a single-phase grounding fault based on multi-source data fusion according to claim 1 is characterized in that: Data preprocessing methods include data cleaning, including: The missing values ​​in the original data set were processed and the mean imputation method was used to estimate and supplement the missing values; Detect and process outliers in the original data set, use the threshold judgment method to identify outliers, and choose to delete outliers, replace them with statistical values, or perform outlier smoothing according to the specific situation; Identify and remove duplicate data in the original dataset.

4. The method for quickly locating a single-phase grounding fault based on multi-source data fusion according to claim 3 is characterized in that: Data preprocessing methods also include: Standardize the continuous data to eliminate the dimension difference. Use the Z-score method to standardize the data. Subtract the mean of each numerical data and divide it by its standard deviation so that the processed data conforms to the standard normal distribution. The formula for standardization is: Z = (X-μ) / σ; where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data; Convert the format and type of the original data set, including converting strings into numeric data, unifying date and time data in different formats into a standard date and time format, and performing one-hot encoding conversion on categorical data.

5. The method for quickly locating a single-phase grounding fault based on multi-source data fusion according to claim 1 is characterized in that: The implementation methods of fault type pattern recognition include: The fault data in the preprocessed single-phase grounding fault basic data set is subjected to feature extraction, and the extracted features include the magnitude of the fault current, the rate of change of the fault voltage, the fault duration, and the system state parameters before and after the fault occurs; The K-means clustering algorithm is used to perform cluster analysis on the extracted features to identify different fault type patterns; the goal of the clustering algorithm is to minimize the intra-class differences and maximize the inter-class differences, so as to accurately classify different fault types.

6. The method for quickly locating a single-phase grounding fault based on multi-source data fusion according to claim 1 is characterized in that: The implementation methods of system operation status impact analysis include: Extracting features from the system operating status data in the preprocessed single-phase grounding fault basic data set, wherein the features include system voltage, current, frequency, load rate, reactive power factor and operating status parameters of key equipment; The Pearson correlation coefficient analysis algorithm is used to calculate the correlation between the extracted system operation status characteristics and the occurrence of single-phase grounding faults. By quantifying the degree of correlation between the characteristics and the faults, the system operation status factors that have a significant impact on the occurrence of the faults are identified. The calculation formula of the Pearson correlation coefficient r is: Among them, r represents the correlation coefficient between the feature and the fault, n represents the number of data points, x represents the feature value, and n represents the probability value of the fault occurrence.

7. The method for quickly locating a single-phase grounding fault based on multi-source data fusion according to claim 1 is characterized in that: The steps to build a dual machine learning model include: A fault prediction model is constructed for the prediction of the occurrence of single-phase grounding faults. The model adopts a logistic regression classification algorithm, with the input being the feature vector of power system operation data, environmental data and historical fault data, and the output being the occurrence probability and possible time period of single-phase grounding faults. The formula of the logistic regression model is: Where P(Y=1|X) represents the probability of a single-phase ground fault under a given input feature vector X; β0,β1,…,β n are model parameters, learned through training data; To determine the location of the single-phase grounding fault, a positioning model is constructed; the model uses the support vector regression SVR algorithm, the input is the power system state data when the fault occurs, the data recorded by the fault recorder and the data of the distributed sensor, and the output is the specific location of the fault; the formula of the SVR model is: Where f(x) represents the predicted value of the fault location; α i and is the Lagrange multiplier, which is learned through training data; K(x i ,x) is the kernel function, which is used to measure the similarity between input feature vectors; b is the bias term.

8. The method for quickly locating a single-phase grounding fault based on multi-source data fusion according to claim 7 is characterized in that: In the process of building the dual machine learning model, cross-validation and gradient descent algorithms are used to optimize the model parameters; cross-validation is used to evaluate the generalization ability of the model and prevent overfitting; the gradient descent algorithm is used to iteratively update the model parameters so that the loss function of the model on the training data reaches the minimum value.

9. A method for quickly locating a single-phase grounding fault based on multi-source data fusion according to claim 8, characterized in that: Develop targeted troubleshooting plans based on the output results of the fault prediction model; If a high risk of single-phase grounding fault is predicted within a certain period of time, the monitoring and inspection of the power system will be strengthened during that period.

10. A single-phase grounding fault rapid location method based on multi-source data fusion according to claim 9, characterized in that: A rapid positioning strategy is developed based on the output results of the positioning model; once a single-phase grounding fault actually occurs, the real-time data at the time of the fault is immediately input into the positioning model to quickly and accurately determine the fault location.

Images