Transformer substation single-end measuring point fault ranging and positioning system

Abstract

The invention discloses a transformer station single-end measuring point fault distance measurement positioning system, and relates to the field of power grids, and the system comprises a data collection module which is used for collecting the multi-dimensional data of voltage, current, temperature, vibration and the like at the key position of a transformer station in real time; the data processing and preprocessing module is used for cleaning, screening and normalizing the collected data and applying a data compression and feature extraction technology; the machine learning model training module is used for constructing a training data set by using historical fault data, selecting a proper machine learning algorithm to carry out model training, and optimizing model parameters through cross validation; the fault mode recognition and prediction module is used for receiving newly collected data in real time, performing fault mode recognition by using a trained machine learning model, and predicting potential fault types and occurrence positions; and the fault distance measurement and positioning module is combined with a single-end traveling wave distance measurement method, utilizes artificial intelligence to optimize a wave head identification and wave velocity estimation process, and automatically calculates the position of a fault point.

Description

Technical Field

[0001] The invention relates to the field of power grids, and in particular to a single-end measuring point fault distance measurement and positioning system for a transformer substation. Background Art

[0002] Traditional substation fault detection methods mainly rely on manual inspections and regular maintenance. Although this method is simple and easy to implement, it is inefficient, has poor real-time performance, and is prone to missed detection. It cannot meet the modern power system's requirements for rapid response and high accuracy in fault detection.

[0003] In addition, there are currently some fault detection systems based on automated equipment. These systems usually use sensors to collect data and make fault judgments through preset algorithms. However, these systems are often limited by the accuracy and scope of data collection, as well as the intelligence level of the algorithms, resulting in the accuracy and real-time performance of fault detection still need to be improved.

[0004] In terms of fault location, existing technologies include positioning methods based on line terminals or fault indicators, and distribution network fault location technologies based on traveling wave theory. Although these methods can improve the accuracy of fault location to a certain extent, there are still some problems. For example, positioning methods based on line terminals or fault indicators require a lot of hardware equipment support, which increases the complexity and cost of the system; and positioning methods based on traveling wave theory may be interfered by factors such as traveling wave reflection and refraction in some cases, affecting the accuracy of positioning. Summary of the invention

[0005] The present application provides a substation single-end measuring point fault distance measurement and positioning system, which is used to solve the technical problems that the existing substation fault detection and processing speed is slow and the fault summary needs to be manually processed.

[0006] In view of the above problems, the present application provides a single-end measuring point fault ranging and positioning system for a substation, including: a data acquisition module, which is used to collect multi-dimensional data such as voltage, current, temperature, vibration, etc. at key locations of the substation in real time; a data processing and preprocessing module, which is used to clean, screen, and normalize the collected data, and apply data compression and feature extraction techniques; a machine learning model training module, which uses historical fault data to build a training data set, selects a suitable machine learning algorithm for model training, and optimizes model parameters through cross-validation; a fault pattern recognition and prediction module, which is used to receive newly collected data in real time, use the trained machine learning model to identify fault patterns, and predict potential fault types and locations; a fault ranging and positioning module, which combines the single-end traveling wave ranging method, uses artificial intelligence to optimize the wave head recognition and wave speed estimation process, and automatically calculates the fault point location; a decision support and alarm module, which is used to automatically generate maintenance suggestions and operation guidelines based on the fault prediction and positioning results, and trigger the alarm mechanism when a serious fault is detected.

[0007] In summary, the present invention mainly has the following beneficial effects:

[0008] 1. The present invention combines high-precision sensor technology to collect multi-dimensional data of substation equipment in real time, providing a rich data source for fault detection and positioning. This overcomes the problems of low efficiency and poor real-time performance of traditional inspection methods, allowing faults to be discovered and handled in the first place.

[0009] 2. Use big data and machine learning technology to store, process and analyze the massive amount of data collected, and dig out the fault characteristics and patterns hidden in the data. Through the trained model, the system can automatically identify and predict faults, realizing automatic identification and prediction of faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 The present invention is a system framework diagram of a substation single-end measuring point fault ranging and locating system. DETAILED DESCRIPTION

[0011] The present application provides a substation single-end measuring point fault distance measurement and positioning system, which is used to solve the technical problems that the existing substation fault detection and processing speed is slow and the fault summary needs to be manually processed.

[0012] After introducing the basic principles of the present application, the technical solutions in the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described here. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0013] Embodiment 1

[0014] A single-end measuring point fault distance measurement and positioning system for a substation, comprising:

[0015] Data acquisition module, used to collect multi-dimensional data such as voltage, current, temperature, vibration, etc. in real time at key locations in the substation;

[0016] The data acquisition module is responsible for receiving and processing various sensor data from key locations of the substation. These data include but are not limited to voltage, current, temperature, vibration, etc., which are important parameters for evaluating the operating status of the substation.

[0017] After conversion and processing, these data can be used to analyze the load, power balance and safe operation of the power grid. Through the temperature sensors installed on the substation equipment, the data acquisition module can obtain the operating temperature of the equipment in real time. It is used to promptly detect abnormal conditions such as overheating of the equipment and prevent potential safety hazards.

[0018] The data acquisition module includes a high-precision sensor unit, which is deployed at key locations in the substation to collect multi-dimensional data in real time; a high-speed communication network unit, which is used to transmit data to the data processing center.

[0019] Optional, data validity determination unit: This unit can determine whether the channel is valid based on the sampling sequence number of the input data, ensuring the accuracy and reliability of the collected data.

[0020] Data processing and preprocessing module, used to clean, filter, and normalize the collected data, and apply data compression and feature extraction technology;

[0021] Data cleaning is the first step in data preprocessing, which aims to correct, delete or replace inaccurate, incomplete, erroneous or duplicate data. For example, spelling errors, illogical values, etc. Duplicate data records should be deleted to ensure the uniqueness of the data.

[0022] Data screening is to filter data according to specific conditions to remove data that does not meet the analysis requirements. For example, data is screened according to the range of the data to remove values ​​that are outside the reasonable range. Data is screened according to the type of data to only retain specific types of data for analysis.

[0023] Among them, normalization is to scale the data so that it falls into a smaller specific range, thereby eliminating the dimensional differences between different features and making them comparable.

[0024] The machine learning model training module uses historical fault data to build a training data set, selects a suitable machine learning algorithm for model training, and optimizes model parameters through cross-validation;

[0025] Choose the appropriate machine learning algorithm based on the nature of the problem and the characteristics of the data. For example, for a classification problem, you can choose algorithms such as decision trees, random forests, support vector machines, or neural networks.

[0026] The training data set D contains n fault cases of different types. Each case is represented as (x_i, y_i), where x_i is the input feature vector and y_i is the corresponding fault type label.

[0027] A machine learning algorithm A, selected from one or more algorithms such as support vector machine, random forest, neural network, etc., is used to train the model M: M = A (D);

[0028] The model parameter optimization method O uses k-fold cross-validation to optimize the model parameters, that is, the data set D is divided into k subsets, and k-1 subsets are used as training data each time, and the remaining subset is used as validation data. This is repeated k times, and a different subset is selected as validation data each time. Finally, the model parameters with the smallest average validation error are selected.

[0029] The optimization objective function can be expressed as:

[0030]

[0031] in, Indicates the use of a subset of training data The trained model, represents the corresponding validation data subset, L represents the loss function, and θ represents the model parameters

[0032] The above steps are mainly to find a set of model parameters that optimizes the model's performance on the data. This method ensures a comprehensive evaluation of the model's performance and improves the model's generalization ability through multiple training and validation.

[0033] Fault pattern recognition and prediction module, which receives newly collected data in real time, uses the trained machine learning model to perform fault pattern recognition, and predicts potential fault types and locations;

[0034] This module can receive new data collected from various sensors, monitoring systems or other data sources in real time. This module can load the model previously trained by the machine learning model training module. This model has been trained and optimized with historical fault data and has the ability to identify fault patterns.

[0035] When newly acquired data is fed into the loaded model, the model analyzes it and identifies whether the current data exhibits a known failure pattern. This process is based on the failure characteristics and patterns learned by the model during the training phase.

[0036] Based on the analysis results of the model, the module can predict the potential fault type. For example, it may identify whether the impending fault of the equipment is a mechanical fault, an electrical fault, or another type of fault. In addition to the fault type, the module can also predict the specific location where the fault may occur. This can help maintenance personnel quickly locate and solve the problem.

[0037] Once the potential failure mode is identified and the type and location of the failure are predicted, the module will output this information in an appropriate form, which may include text reports, graphical displays or alarm signals. If the predicted failure is of high severity or requires immediate attention, the module can trigger an alarm mechanism to notify relevant personnel to take necessary maintenance measures in a timely manner.

[0038] The fault distance measurement and location module combines the single-ended traveling wave distance measurement method and uses artificial intelligence to optimize the wave head identification and wave speed estimation process to automatically calculate the fault point location;

[0039] When a power line fails, traveling waves will be generated at the fault point and propagate to both ends of the line. Single-ended traveling wave ranging is to determine the location of the fault point by analyzing the propagation time, reflection and refraction characteristics of these traveling waves. This module first collects these traveling wave signals, which is usually done through highly sensitive sensors or measuring equipment to ensure that weak traveling wave signals are captured.

[0040] Once the wave head and wave speed are determined, the module will use this information and the time difference of the traveling wave from the fault point to the measurement point to automatically calculate the location of the fault point. Through precise time synchronization and high-speed data combined with wave head identification and wave speed estimation, high-precision positioning of the fault point can be achieved.

[0041] It is worth mentioning that after calculating the location of the fault point, the module will output the results in the form of graphics or text, which is convenient for operation and maintenance personnel to quickly understand and respond. At the same time, the module can also trigger the alarm system to promptly notify relevant personnel of the exact location of the fault point so that repairs can be carried out quickly.

[0042] The wave head identification submodule uses advanced signal processing techniques, such as wavelet transform or Hilbert-Huang transform, to accurately identify the arrival time of the traveling wave;

[0043] Wave velocity estimation submodule: This module uses historical fault data and combines machine learning algorithms to estimate the propagation velocity v of traveling waves on the transmission line. The estimation formula can be expressed as:

[0044] v=f(T,C,…)

[0045] Where T is the ambient temperature, C is the physical properties of the conductor, including factors such as material and diameter, and f is the wave velocity calculation function;

[0046] The fault point location calculation submodule calculates the distance d between the fault point and the measurement point using the formula according to the traveling wave arrival time t1 and the reflected wave arrival time t2 determined by the wave head identification submodule. The specific formula is as follows:

[0047]

[0048] Among them, v is the traveling wave velocity obtained by the wave velocity estimation submodule.

[0049] The fault distance measurement and location module also includes a feedback optimization module, which is specifically used to correct the prediction of the fault mode recognition and prediction module according to the feedback of each maintenance result, and continuously optimize it. The specific implementation steps include:

[0050] Collect maintenance feedback data. After each maintenance, collect data about the maintenance results, including maintenance measures, maintenance time, maintenance cost, and equipment status after maintenance, recorded as data set FB = {(mi, ci, si, ti)}, where mi represents maintenance measures, ci represents maintenance cost, si represents equipment status after maintenance, and ti represents maintenance time;

[0051] Evaluate the prediction accuracy, compare the equipment status si in the maintenance feedback data with the prediction results of the fault mode recognition and prediction module, and calculate the prediction accuracy P acc , the formula is:

[0052]

[0053] Where n is the number of maintenance times, δ(si, s^i) is the judgment function, when si is consistent with the predicted result s^i, δ(si, s^i) = 1, otherwise it is 0;

[0054] Update the model parameters according to the prediction accuracy P acc And maintenance feedback data, using gradient descent or other optimization algorithms, the model parameters of the fault mode recognition and prediction module are updated, and the optimization objective function L can be expressed as:

[0055]

[0056] in, Given maintenance measures, maintenance costs, maintenance time and model parameters Under the condition of , the probability that the device state is si, are the model parameters to be optimized;

[0057] Iterate the loop until the prediction accuracy reaches P acc When the preset threshold is reached or the number of iterations reaches the upper limit, the optimization and correction of the model parameters are completed.

[0058] The main function of the feedback and optimization module is to continuously optimize the prediction accuracy of the fault pattern recognition and prediction module based on the feedback of each maintenance result. In actual applications, after performing the maintenance task, the maintenance personnel will feedback the maintenance results to the system. The feedback and optimization module will receive this feedback information and compare it with the previous prediction results.

[0059] The maintenance results referred to herein include maintenance measures, replaced parts, equipment status after maintenance, etc. The types of available reference information may increase or decrease depending on the actual situation.

[0060] In order to quantify the accuracy of the prediction, the feedback and optimization module can use a series of evaluation indicators, such as accuracy, recall, F1 score, etc. These evaluation indicators are calculated based on the comparison between the actual repair results and the predicted results. For example, the accuracy rate can be defined as the ratio of the number of correctly predicted faults to the total number of predicted faults.

[0061] Based on these evaluation indicators, the feedback and optimization module will adjust and optimize the model parameters of the fault pattern recognition and prediction module. This can be achieved through a variety of methods, such as gradient descent, stochastic gradient descent, genetic algorithm, etc. These optimization methods aim to minimize the prediction error and improve the prediction accuracy of the model.

[0062] It is worth mentioning that the feedback and optimization module can also use historical maintenance data to build a maintenance case library. This case library will record detailed information for each maintenance, including fault type, maintenance measures, maintenance results, etc. When a new maintenance case occurs, the module can match it with the historical cases in the case library to find the most similar historical cases as a reference. This not only improves maintenance efficiency, but also provides valuable data for future fault prediction and maintenance. Optionally, the case library is connected to the cloud and classified according to the installation and use environment. Fault data under the same use environment has a certain reference, and the data can be interoperable to improve the efficiency of optimization and help new equipment achieve optimization goals better and faster.

[0063] To sum up, by introducing the feedback and optimization module, the substation single-end measuring point fault ranging and positioning system described in this application can achieve self-learning and optimization, continuously improve the accuracy of fault prediction and positioning, and provide more reliable guarantee for the safe and stable operation of the power system.

[0064] Please note that the above specific implementation methods are partly reasonably speculated and constructed based on the content of previously uploaded documents and common technical practices. In actual applications, the specific implementation methods may be adjusted and optimized according to actual needs and technical conditions.

[0065] The decision support and alarm module is used to automatically generate maintenance suggestions and operation guidelines based on the fault prediction and location results, and trigger the alarm mechanism when a serious fault is detected.

[0066] When the fault pattern recognition and prediction module predicts the potential fault type and location, the decision support and alarm module will automatically generate corresponding maintenance suggestions based on preset rules and algorithms. These suggestions may include replacing specific parts, cleaning or lubricating operations, adjusting equipment parameters, etc., to help maintenance personnel solve problems quickly and effectively.

[0067] In addition to maintenance recommendations, the module can also generate detailed operating instructions. These instructions may include maintenance steps, a list of required tools, safety precautions, etc. to ensure the safety and efficiency of the maintenance process. This has the effect of standardizing maintenance operations.

[0068] When the fault location module determines the exact location of the fault point, the decision support and alarm module can use this information to provide more precise guidance and suggestions to maintenance personnel. For example, it can point out the equipment part that needs to be checked in detail, or provide the best path to the fault point.

[0069] The module continuously monitors the received fault prediction and location data. Once a serious fault is detected, that is, those that may pose an immediate threat to the safety of equipment or personnel, the module will immediately trigger an alarm mechanism. The alarm may include various forms such as sound alarms, flashing lights, SMS or email notifications to ensure that relevant personnel can be notified quickly and take action. At the same time, the alarm information will also contain detailed information about the fault, such as fault type, location and preliminary repair suggestions.

[0070] Optionally, by collecting and analyzing feedback from maintenance personnel and actual maintenance data of the equipment, the module can continuously optimize the maintenance suggestions and operation guidelines it generates, thereby improving maintenance efficiency and success rate.

[0071] It is worth mentioning that this application also includes a visual interface that is compatible with the system, which is used to display fault prediction and location results, as well as maintenance suggestions and operation guides; and an alarm mechanism, including one or more alarm methods such as sound alarm, light alarm or SMS alarm.

[0072] Among them, the feature selection and extraction algorithm is used to extract the most useful feature set F for fault identification from the original data;

[0073] The feature dimension reduction method uses principal component analysis PCA, linear discriminant analysis LDA and other methods to reduce the feature set F to a lower dimension F'.

[0074] The substation single-end measurement point fault distance measurement and positioning system according to any of the preceding claims, characterized in that the machine learning model also includes:

[0075] Ensemble learning methods use algorithms such as random forest RF and gradient boosted tree GBDT to improve the generalization ability and stability of the model;

[0076] Deep learning models, including convolutional neural networks (CNNs) and long short-term memory (LSTMs), are used to process time series data and capture complex failure modes.

[0077] The substation single-end measuring point fault ranging and positioning system according to claim 1 further comprises:

[0078] The network integration module of the power grid monitoring system enables data sharing and collaborative work;

[0079] The mobile application APP signal connection module allows operators to view fault information and perform maintenance tasks anytime and anywhere.

[0080] Those skilled in the art will understand that the various digital numbers such as the first and second involved in this application are only for the convenience of description, and are not used to limit the scope of this application, nor do they indicate a sequence of precedence. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated with each other are in an "or" relationship. "At least one" refers to one or more. At least two refers to two or more. "At least one", "any one" or similar expressions refer to any combination of these items, including any combination of single items (individuals) or plural items (individuals). For example, at least one item (individual, kind) of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0081] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in this application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0082] The steps of the method or algorithm described in this application can be directly embedded in hardware, software units executed by a processor, or a combination of the two. The software unit can be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, register, hard disk, removable disk, CD-ROM or any other form of storage medium in the art. Exemplarily, the storage medium can be connected to the processor so that the processor can read information from the storage medium and write information to the storage medium. Optionally, the storage medium can also be integrated into the processor. The processor and the storage medium can be arranged in an ASIC, and the ASIC can be arranged in a terminal. Optionally, the processor and the storage medium can also be arranged in different components in the terminal. These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are performed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0083] Although the present application has been described in conjunction with specific features and embodiments thereof, it is obvious that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present application and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A single-ended measuring point fault distance measurement and positioning system for a substation, characterized in that: include: Data acquisition module, used to collect multi-dimensional data such as voltage, current, temperature, vibration, etc. in real time at key locations in the substation; Data processing and preprocessing module, used to clean, filter, and normalize the collected data, and apply data compression and feature extraction technology; The machine learning model training module uses historical fault data to build a training data set, selects a suitable machine learning algorithm for model training, and optimizes model parameters through cross-validation; Fault pattern recognition and prediction module, which receives newly collected data in real time, uses the trained machine learning model to perform fault pattern recognition, and predicts potential fault types and locations; The fault distance measurement and location module combines the single-ended traveling wave distance measurement method and uses artificial intelligence to optimize the wave head identification and wave speed estimation process to automatically calculate the fault point location; The decision support and alarm module is used to automatically generate maintenance suggestions and operation guidelines based on the fault prediction and location results, and trigger the alarm mechanism when a serious fault is detected.

2. The substation single-end measuring point fault ranging and positioning system according to claim 1, characterized in that: The data acquisition module comprises: High-precision sensor units are deployed at key locations in substations to collect multi-dimensional data in real time; High-speed communication network unit used to transmit data to the data processing center.

3. The substation single-end measuring point fault ranging and positioning system according to claim 1, characterized in that: The machine learning model training module includes: The training data set D contains n fault cases of different types. Each case is represented as (x_i, y_i), where x_i is the input feature vector and y_i is the corresponding fault type label. A machine learning algorithm A, selected from one or more algorithms such as support vector machine, random forest, neural network, etc., is used to train the model M: M = A (D); The model parameter optimization method O uses k-fold cross-validation to optimize the model parameters, that is, the data set D is divided into k subsets, and k-1 subsets are used as training data each time, and the remaining subset is used as validation data. This is repeated k times, and a different subset is selected as validation data each time. Finally, the model parameters with the smallest average validation error are selected. The optimization objective function can be expressed as: in, Indicates the use of a subset of training data The trained model, represents the corresponding validation data subset, L represents the loss function, and θ represents the model parameters.

4. The substation single-end measuring point fault ranging and positioning system according to claim 1, characterized in that: The fault distance measurement and positioning module includes: The wave head identification submodule uses advanced signal processing techniques, such as wavelet transform or Hilbert-Huang transform, to accurately identify the arrival time of the traveling wave; Wave velocity estimation submodule: This module uses historical fault data and combines machine learning algorithms to estimate the propagation velocity v of traveling waves on the transmission line. The estimation formula can be expressed as: v=f(T,C,…) Where T is the ambient temperature, C is the physical properties of the conductor, including factors such as material and diameter, and f is the wave velocity calculation function; The fault point location calculation submodule calculates the distance d between the fault point and the measurement point using the formula according to the traveling wave arrival time t1 and the reflected wave arrival time t2 determined by the wave head identification submodule. The specific formula is as follows: Among them, v is the traveling wave velocity obtained by the wave velocity estimation submodule.

5. The substation single-end measuring point fault ranging and positioning system according to claim 4, characterized in that: The fault distance measurement and positioning module includes: It also includes a feedback optimization module, which is specifically used to correct the prediction of the fault mode recognition and prediction module according to the feedback of each maintenance result, and continuously optimize it. The specific implementation steps include: Collect maintenance feedback data. After each maintenance, collect data about the maintenance results, including maintenance measures, maintenance time, maintenance cost, and equipment status after maintenance, recorded as data set FB = {(mi, ci, si, ti)}, where mi represents maintenance measures, ci represents maintenance cost, si represents equipment status after maintenance, and ti represents maintenance time; Evaluate the prediction accuracy, compare the equipment status si in the maintenance feedback data with the prediction results of the fault mode recognition and prediction module, and calculate the prediction accuracy P acc , the formula is: Where n is the number of maintenance times, δ(si, s^i) is the judgment function, when si is consistent with the predicted result s^i, δ(si, s^i) = 1, otherwise it is 0; Update the model parameters according to the prediction accuracy P acc And maintenance feedback data, using gradient descent or other optimization algorithms, the model parameters of the fault mode recognition and prediction module are updated, and the optimization objective function L can be expressed as: in, Given maintenance measures, maintenance costs, maintenance time and model parameters Under the condition of , the probability that the device state is si, are the model parameters to be optimized; Iterate the loop until the prediction accuracy reaches P acc When the preset threshold is reached or the number of iterations reaches the upper limit, the optimization and correction of the model parameters are completed.

6. The substation single-end measuring point fault distance measurement and positioning system according to claim 1, further comprising: Visual interface for displaying fault prediction and location results, as well as maintenance suggestions and operation guides; The alarm mechanism includes one or more alarm modes such as sound alarm, light alarm or text message alarm.

7. The substation single-end measuring point fault distance measurement and positioning system according to claim 1, further comprising: Feature selection and extraction algorithm, used to extract the most useful feature set F for fault identification from the original data; The feature dimension reduction method uses principal component analysis PCA, linear discriminant analysis LDA and other methods to reduce the feature set F to a lower dimension F'.

8. The substation single-end measuring point fault distance measurement and positioning system according to any of the preceding claims, characterized in that: The machine learning model also includes: Ensemble learning methods use algorithms such as random forest RF and gradient boosted tree GBDT to improve the generalization ability and stability of the model; Deep learning models, including convolutional neural networks (CNNs) and long short-term memory (LSTMs), are used to process time series data and capture complex failure modes.

9. The substation single-end measuring point fault distance measurement and positioning system according to claim 1, further comprising: The network integration module of the power grid monitoring system enables data sharing and collaborative work; The mobile application APP signal connection module allows operators to view fault information and perform maintenance tasks anytime and anywhere.

Images