Switch cabinet latent fault remote diagnosis system, method, device and medium

Abstract

The invention discloses a switch cabinet latent fault remote diagnosis system, method, device and medium, and belongs to the technical field of fault diagnosis, and the system comprises a multi-dimensional sensor sensing module, a data transmission module, a data fusion analysis module, and a data storage and fault detection module. The multi-dimensional sensor sensing module is used for acquiring multi-dimensional data of the operation state of the switch cabinet; the data transmission module is used for transmitting the multi-dimensional data to the data fusion analysis module; the data fusion analysis module is used for performing data preprocessing on the multi-dimensional data to obtain preprocessed multi-dimensional data, and determining a fault classification result according to the preprocessed multi-dimensional data; and the data storage and fault detection module is used for performing fault identification based on the real-time monitoring data and the fault classification result. According to the invention, early recognition and accurate diagnosis of the latent fault of the switch cabinet are realized, and the downtime and maintenance cost of equipment are reduced.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to a remote diagnosis system, method, device and medium for latent faults in switchgear. Background Technology

[0002] Air circuit breakers (ACBs) are an important component of power systems used to protect electrical equipment. They are widely used in industrial, commercial, and building sectors. The main functions and applications of air circuit breakers include overload protection, short-circuit protection, current monitoring and control, and prevention of electrical fires. They are used to ensure the safe operation of power systems, extend the service life of equipment, and provide fault isolation functions.

[0003] However, existing technologies rely on simple sensor data to determine whether partial discharge has occurred. Although they consider factors such as gas composition, temperature, humidity, and particle density, the analysis logic is relatively linear and lacks the ability to cross-analyze data from multiple sensors. Secondly, discharge phenomena inside gas switchgear vary in intensity, and different devices may exhibit different discharge modes. Existing technologies use a uniform threshold for judgment, which may lead to oversensitivity or insensitivity, resulting in false alarms or missed alarms. Furthermore, while existing technologies can detect partial discharge, they lack remote monitoring, data analysis, and decision support functions. For large power companies or grid operators, real-time monitoring and intelligent decision support systems are crucial. Therefore, the development and research of a remote comprehensive diagnostic system for latent faults in switchgear based on multi-sensor information fusion analysis has become a key concern for ensuring power system safety. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by this invention are: how to address the issue that air switchgear may fail to disconnect circuits in time when encountering overload or short circuit, potentially damaging electrical equipment; simultaneously, the internal temperature of the air switchgear may rise due to poor ventilation, excessive load, or other reasons, and excessively high temperatures may damage internal components and cause fires; faults or leaks in the gas system can lead to serious safety problems, and temperature and humidity safety detection directly affects the stability of the equipment. The invention utilizes an effective gas and temperature / humidity monitoring system to ensure the equipment operates in a safe and stable state, providing timely warnings of potential faults and preventing safety accidents; and during operation, to address the potential for faults such as arc discharge, poor contact, overcurrent, and insulation aging, which are often accompanied by abnormal changes in the electromagnetic field, the invention utilizes electromagnetic inductive sensors in the air switchgear to improve the real-time performance of fault detection, reduce sudden accidents, and achieve intelligent operation and maintenance.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a remote diagnostic system for latent faults in switchgear, comprising the following steps,

[0007] The system comprises a multi-dimensional sensor sensing module, a data transmission module, a data fusion and analysis module, and a data storage and fault detection module. The multi-dimensional sensor sensing module acquires multi-dimensional data on the switchgear's operating status. The data transmission module transmits the multi-dimensional data to the data fusion and analysis module. The data fusion and analysis module preprocesses the multi-dimensional data to obtain preprocessed multi-dimensional data and determines fault classification results based on the preprocessed multi-dimensional data. The data storage and fault detection module identifies faults based on real-time monitoring data and the fault classification results.

[0008] As a preferred embodiment of the remote diagnostic system for latent faults in switchgear according to the present invention, the multidimensional data is preprocessed, including: performing time synchronization processing on the multidimensional data to obtain time-synchronized multidimensional data; resampling the time-synchronized multidimensional data to obtain resampled multidimensional data; performing denoising and normalization processing on the resampled multidimensional data; and detecting and interpolating abnormal data. The beneficial effect of this preferred embodiment is that by employing a complete preprocessing flow of time synchronization, resampling, denoising, and normalization, it can effectively eliminate time deviations and dimensional differences between multidimensional sensor data, improving data quality and the accuracy of subsequent fault classification.

[0009] As a preferred embodiment of the remote diagnostic system for latent faults in switchgear according to the present invention, the method for determining the fault classification result based on the preprocessed multidimensional data includes: extracting data features from the preprocessed multidimensional data to determine feature data; performing multi-source data fusion based on the feature data to determine a weighted fused feature vector; and performing fault classification based on the weighted fused feature vector to obtain the fault classification result.

[0010] As a preferred embodiment of the remote diagnostic system for latent faults in switchgear described in this invention, the fault identification based on real-time monitoring data and the fault classification results includes: classifying the abnormal state of the switchgear according to the level of the abnormal state based on the real-time monitoring data and the fault classification results; judging the threshold based on the level of abnormal state; and marking and warning the equipment area according to the threshold judgment result.

[0011] As a preferred embodiment of the remote diagnostic system for latent faults in switchgear according to the present invention, the following steps are performed: data feature extraction is performed on the preprocessed multidimensional data to determine the feature data, including: local time-series feature extraction is performed on the preprocessed multidimensional data to obtain time-domain features, frequency-domain features and time-series features.

[0012] As a preferred embodiment of the remote diagnostic system for latent faults in switchgear according to the present invention, the system involves: fusing multi-source data based on the characteristic data to determine a weighted fused feature vector, including: calculating the importance of the time-domain features, frequency-domain features, and time-series features; and, based on the importance, performing feature concatenation and weighted fusion on the time-domain features, frequency-domain features, and time-series features to obtain the weighted fused feature vector. The beneficial effect of this preferred embodiment is that by calculating the importance of time-domain features, frequency-domain features, and time-series features and performing weighted fusion, the complementary advantages of various features can be fully utilized, effectively improving the expressive power of fault features and the robustness of fault identification.

[0013] In a preferred embodiment of the remote diagnostic system for latent faults in switchgear described in this invention, the multi-dimensional sensor sensing module comprises: a particle density sensor, a temperature and humidity sensor, a gas content sensor, and an electromagnetic radiation sensor arranged in a specific order within a gas pipeline; and an air pump placed within the gas pipeline to draw in external gas for sampling, thereby sensing the internal conditions of the switchgear. The beneficial effect of this preferred embodiment is that by arranging four types of sensors in a specific order within the gas pipeline and configuring an air pump for active sampling, comprehensive monitoring of the internal environment of the switchgear can be achieved, improving the early detection capability and detection coverage of latent faults.

[0014] Another objective of this invention is to provide a remote diagnostic method for latent faults in switchgear.

[0015] To address the aforementioned technical problems, this invention provides the following technical solution: a remote diagnosis method for latent faults in switchgear, comprising: acquiring multidimensional data on the operating status of the switchgear using a multidimensional sensor sensing module; transmitting the multidimensional data to a data fusion analysis module using a data transmission module; preprocessing the multidimensional data using the data fusion analysis module to obtain preprocessed multidimensional data; determining fault classification results based on the preprocessed multidimensional data; and identifying faults based on real-time monitoring data and fault classification results using a data storage and fault detection module.

[0016] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the remote diagnosis system for latent faults in switchgear.

[0017] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements the steps of the remote diagnostic system for latent faults in switchgear.

[0018] The beneficial effects of this invention are as follows: By deploying multiple sensors within gas pipelines, this invention comprehensively captures the operating status information of switchgear. Real-time data transmission is achieved using Wi-Fi wireless communication technology. Data preprocessing, feature extraction, and multi-source data fusion improve data quality and information utilization. Accurate fault classification is achieved based on support vector machines. Finally, the results are displayed through a graphical interface, providing decision support. This invention solves the problems of insufficient information dimensions, poor data quality, and low diagnostic accuracy in traditional switchgear monitoring systems. It can capture abnormal signals in the early stages of faults, preventing problems before they occur. In particular, the application of multi-sensor fusion technology and machine learning methods improves the ability to identify various latent faults, avoiding missed or false diagnoses caused by single-parameter monitoring. Through multi-sensor data fusion and intelligent analysis, early identification and accurate diagnosis of latent faults in switchgear are achieved, improving the accuracy and timeliness of fault diagnosis, reducing equipment downtime and maintenance costs, extending the service life of switchgear equipment, and simultaneously enhancing the safety and reliability of power system operation. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0020] Figure 1 The present invention provides a flowchart of a remote diagnostic system for latent faults in switchgear, as an embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of the structure of a multi-dimensional sensor sensing module provided in one embodiment of the present invention.

[0022] Figure 3 This is a flowchart of multi-source data processing provided in one embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram of the structure of a computer device provided in one embodiment of the present invention. Detailed Implementation

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0026] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0027] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a remote diagnostic system for latent faults in switchgear, comprising:

[0028] The system includes a multi-dimensional sensor perception module, a data transmission module, a data fusion and analysis module, and a data storage and fault detection module.

[0029] In this embodiment of the invention, the multi-dimensional sensor sensing module is used to acquire multi-dimensional data on the operating status of the switchgear.

[0030] In this embodiment of the invention, the data transmission module is used to transmit multidimensional data to the data fusion and analysis module.

[0031] In this embodiment of the invention, the data fusion analysis module is used to preprocess the multidimensional data to obtain preprocessed multidimensional data, and to determine the fault classification result based on the preprocessed multidimensional data.

[0032] In this embodiment of the invention, the data storage and fault detection module is used to identify faults based on real-time monitoring data and fault classification results.

[0033] It should be noted that traditional switchgear fault diagnosis systems often rely on single sensors to monitor local parameters, making it difficult to comprehensively reflect the equipment's operating status. Sensor data is susceptible to environmental factors such as electromagnetic interference and temperature and humidity fluctuations, resulting in unstable data quality. Fault feature extraction methods often employ simple threshold judgments, failing to effectively identify the early, subtle characteristics of latent faults. Existing systems lack the ability to fuse and analyze multi-source data, leading to low fault classification accuracy and high false alarm rates. Furthermore, there is a disconnect between equipment status monitoring data and maintenance decisions, making predictive maintenance difficult. These issues result in deficiencies in the early detection and accurate location of latent faults in switchgear.

[0034] Therefore, to address the aforementioned monitoring blind spots and diagnostic accuracy issues, a multi-dimensional sensor sensing module, data transmission module, data fusion and analysis module, and data storage and fault detection module are implemented. Specifically, a multi-dimensional sensing module integrating particle density, temperature and humidity, gas content, and electromagnetic radiation sensors is deployed. Gas circulation sampling is driven by an air pump to achieve comprehensive synchronous monitoring of environmental parameters within the switchgear. A data transmission channel is established based on the Wi-Fi communication protocol, and NTP clock synchronization technology ensures unified timestamps for multi-source data. Data cleaning is performed using Kalman filtering and the sliding window mean method. Fourier transform is used to extract electromagnetic radiation frequency domain features, and a time-series feature prediction model is constructed using an LSTM network. Entropy weighting is used to determine the feature weights of each sensor and perform weighted fusion. A support vector machine classification model based on a multinomial kernel function is established, and hyperparameters are optimized through grid search to achieve accurate fault classification.

[0035] Example 2, refer to Figure 1 and Figure 2 This is the second embodiment of the present invention. Based on the above embodiments, a remote diagnostic system for latent faults in switchgear is provided.

[0036] In embodiments of the present invention, such as Figure 2 The diagram shows the structure of a multi-dimensional sensor sensing module. Constructing the multi-dimensional sensor sensing module includes the following steps A1-A2:

[0037] A1: Arrange the particle density sensor, temperature and humidity sensor, gas content sensor and electromagnetic radiation sensor in the gas pipeline according to the arrangement order, from front to back.

[0038] A2: Place the air pump inside the gas pipeline to draw in external gas for sampling and to sense the internal condition of the switch cabinet.

[0039] In this embodiment of the invention, obtaining multidimensional data on the operating status of the switchgear includes the following steps B1-B5:

[0040] B1: Calculate the gas pressure based on the cabinet volume, gas leakage, and initial gas concentration.

[0041] B2: Calculate temperature and humidity based on convective heat transfer coefficient, contact area, object temperature, and ambient temperature.

[0042] B3: Calculate gas content using gas pressure.

[0043] B4: Calculate the electric field strength based on Gauss's law and by combining charge density and the electric constant of free space.

[0044] B5: Calculate the magnetic field strength based on Ampere's circuital law and by combining the changes in current density and electric displacement field with time.

[0045] For example, the calculation of gas pressure in step B1 can be represented by the following formula:

[0046]

[0047] Where P is the gas pressure, in Pa; n is the number of moles of gas, in mol; R is the gas constant, in J / mol·K; T is the temperature, in K; and V is the volume, in m³. 3 .

[0048] Specifically, the temperature in step B2 is expressed in the following form:

[0049]

[0050] Where h is the heat dissipation coefficient, in W / m²·K; A is the heat dissipation surface area, in m²; T env Ambient temperature, in K; Q is the total heat generated inside the air switch cabinet, in W.

[0051] Specifically, the humidity in step B2 is expressed as follows:

[0052]

[0053] Where RH represents relative humidity; P vapor P represents the actual water vapor pressure, in Pa. sat This is the saturated water vapor pressure, expressed in Pa.

[0054] Specifically, the gas content in step B3 is expressed as follows:

[0055]

[0056] Among them, C Q P represents gas concentration (mol / m3); Q P represents a partial pressure (Pa) of gases such as CO, NO2, NO, and O3; total The total gas pressure is expressed in Pa.

[0057] Specifically, the electric field strength in step B4 is expressed in the following form:

[0058]

[0059] Where E is the electric field strength (V / m); ρ is the charge density (C / m³); and ε₀ is the electric constant of free space (8.854 × 10⁻⁶). -12 F / m); This is the gradient operator.

[0060] Specifically, the magnetic field strength in step B5 is expressed in the following form:

[0061]

[0062] Where H is the magnetic field strength (A / m); J is the current density (A / m2); The term representing the change of electric displacement field with time (C / m) 2 ·s).

[0063] In one alternative implementation, multidimensional data on the operating status of the switchgear can also be obtained using edge computing. Data preprocessing and feature extraction are performed on sensor nodes, and gas pressure, temperature and humidity, gas content, and electromagnetic field strength are directly calculated through embedded algorithms, thereby reducing data transmission volume and improving real-time performance.

[0064] In another alternative implementation, multidimensional data on the operating status of the switchgear can also be obtained using distributed sensing network technology. Multiple micro-sensor nodes are deployed to form a grid-like monitoring array. Each node works collaboratively through a wireless self-organizing network. The location of the anomaly source is determined using the triangulation principle, and the multidimensional data is integrated through a Bayesian fusion algorithm to improve the accuracy and reliability of fault location.

[0065] It should be noted that by sequentially arranging particle density sensors, temperature and humidity sensors, gas content sensors, and electromagnetic radiation sensors within the gas pipeline, and by setting up an air pump to draw external gas into the pipeline for sampling, the multi-dimensional sensor sensing module achieves multi-dimensional and comprehensive monitoring of the switchgear's internal state. This allows the sensors to operate systematically within the closed pipeline, avoiding the information dimension deficiencies of traditional single-sensor monitoring methods. In particular, the addition of electromagnetic radiation detection compensates for the common neglect of the impact of electromagnetic anomalies on latent faults in switchgear in existing technologies. Secondly, the collaborative operation of the multi-dimensional sensors enables comprehensive perception of the internal gas composition, temperature and humidity environment, and electromagnetic state of the switchgear, effectively avoiding missed and false judgments caused by single-parameter monitoring. This enhances the ability to identify early latent faults in the switchgear, providing comprehensive raw data for subsequent data analysis and fault diagnosis, thereby improving the accuracy and reliability of monitoring, effectively extending the service life of the switchgear, and reducing safety hazards and economic losses caused by sudden failures.

[0066] In this embodiment of the invention, the data transmission module uses Wi-Fi wireless communication technology to remotely transmit the multi-dimensional data of the switch cabinet's operating status to the data processing system in real time. The data transmission module is configured with Raspberry Pi + USB Wi-Fi, connects to the local Wi-Fi network, and publishes data through a communication protocol.

[0067] It should be noted that during data preprocessing and cleaning, multi-sensor data often contains noise, missing values, or inconsistencies, requiring data preprocessing and cleaning to ensure data quality. For data such as current, voltage, and gas concentration from sensors in air switch cabinets, machine learning algorithms (such as Isolation Forest) are used to remove abnormal readings to avoid affecting the system's fusion analysis results. When temperature and humidity sensors fail for a short period of time, interpolation methods are used to calculate the missing data to ensure that the system analysis is not interrupted due to the loss of some data.

[0068] In this embodiment of the invention, data preprocessing of multidimensional data includes the following steps C1-C3:

[0069] C1: Perform time synchronization processing on the multidimensional data to obtain the time-synchronized multidimensional data.

[0070] C2: Resample the multidimensional data after time synchronization to obtain resampled multidimensional data.

[0071] C3: Perform denoising and normalization on the resampled multidimensional data, and detect and impute outlier data.

[0072] For example, during data preprocessing, a unified clock source is used to ensure the consistency of timestamps for the multi-dimensional data of the switchgear's operating status, such as the NTP protocol; a sliding window mean method is used to uniformly resample the multi-dimensional data of the switchgear's operating status; Kalman filtering is used to denoise the data, and the multi-dimensional data of the switchgear's operating status is normalized to detect and interpolate abnormal data.

[0073] In this embodiment of the invention, determining the fault classification result based on the preprocessed multidimensional data includes the following steps D1-D3:

[0074] D1: Extract data features from the preprocessed multidimensional data to determine the feature data.

[0075] D2: Perform multi-source data fusion based on feature data to determine the weighted fused feature vector.

[0076] D3: Fault classification is performed based on the weighted fused feature vectors to obtain the fault classification results.

[0077] It should be noted that when extracting and selecting data features, since electromagnetic radiation anomalies are usually manifested as an increase in energy in certain specific frequency bands of the spectrum, the frequency domain features are obtained by performing Fourier transform (FFT) on the time series data such as the mean, standard deviation, peak value, skewness, and kurtosis of the time series data. Secondly, for the extracted time series features, local time series features are extracted using a sliding window, and future faults are further predicted using the time series analysis method (LSTM).

[0078] Specifically, step D1 involves extracting data features from the preprocessed multidimensional data to determine the feature data, including the following step D11:

[0079] D11: Local time-series feature extraction is performed on the preprocessed multidimensional data to obtain time-domain features, frequency-domain features, and time-series features.

[0080] For example, obtaining time-domain features, frequency-domain features, and time-series features in step D11 may specifically include:

[0081] Extract the mean and standard deviation of temperature and humidity based on temperature and humidity.

[0082] The data fluctuation of gas content is extracted based on gas content.

[0083] Energy characteristics of electromagnetic data are extracted based on electric field strength and magnetic field strength.

[0084] In one optional implementation, data feature extraction is performed on the preprocessed multidimensional data to determine the feature data. Wavelet transform technology can also be used to perform time-frequency analysis on the multidimensional data to extract feature information at different scales, capture the transient and persistent change features of switchgear faults, and effectively identify the precursor signals of latent faults.

[0085] In another alternative implementation, data feature extraction is performed on the preprocessed multidimensional data to determine the feature data. A deep autoencoder network can also be applied to automatically extract high-order nonlinear features from the original sensor data through unsupervised learning, thus eliminating the need for manual feature design. It can adaptively learn the inherent representation structure of the data and detect abnormal patterns by reconstructing errors. This is suitable for handling the correlation of multi-source heterogeneous sensor data in switch cabinets, improving the accuracy and robustness of feature extraction.

[0086] Specifically, step D2 involves multi-source data fusion based on feature data to determine the weighted fused feature vector, including the following steps D21-D22:

[0087] D21: The importance of calculating time-domain features, frequency-domain features, and time-series features.

[0088] D22: Based on importance, the time-domain features, frequency-domain features, and time-series features are concatenated and weighted to obtain a weighted fused feature vector.

[0089] For example, the importance of calculating time-domain features, frequency-domain features, and time-series features in step D21 can be specifically achieved through the following steps:

[0090] Machine learning algorithms are applied to diagnose and classify faults based on the extracted time-domain features, frequency-domain features, and time-series features. The number of sensors is N, and the observations provided by each sensor i are as follows:

[0091] X i (t)={x i (1) (t),x i (2) (t),...,x i (ki) (t)};

[0092] Among them, X i (t) is a sequence consisting of all sensor observations; x i (ki) (t) represents the observation value of each sensor; ki represents the number of features extracted by sensor i; and t represents the timestamp.

[0093] The data from each sensor is standardized to eliminate differences in the units and magnitudes of data from different sensors; it is assumed that for each sensor i, its standardized feature data is expressed as:

[0094]

[0095] in, For standardized results; μ i (j) and σ i j Let be the mean and standard deviation of the j-th feature of sensor i, respectively.

[0096] For each time t, the data from all sensors are concatenated to form a unified feature vector f(t). Assuming the i-th sensor has ki features, the specific formula for the concatenated feature vector is as follows:

[0097]

[0098] Where f(t) is the feature vector after concatenating the data features of all sensors at time t, and the dimension of the feature vector is k1+k2+…+kN.

[0099] Since different sensors may have different impacts on fault diagnosis and prediction, different weights can be assigned to the features of different sensors; where the weight vector is w=[ω1,ω2,...,ω N The specific formula for the weighted fused feature vector is as follows:

[0100]

[0101] Among them, f weight (t) represents the weighted fused feature vector; ω N Let be the weight value of the Nth sensor.

[0102] It should be noted that the weights are set according to the importance and reliability of the sensors. The weighted and fused feature vectors are used as input to machine learning algorithms to train classifiers or regression models for tasks such as fault diagnosis, condition monitoring, and fault prediction. Secondly, the main purpose of multi-sensor data fusion is to integrate data from multiple sensors to improve the reliability, accuracy, and completeness of the data, thereby enhancing the system's fault diagnosis capabilities.

[0103] It should be noted that by implementing data preprocessing, feature extraction, and multi-source data fusion steps, the problems of inconsistent timing, noise interference, and heterogeneity in multi-sensor data were solved. First, a unified clock source was used to ensure consistent timestamps across all sensor data, laying the foundation for subsequent data fusion. Second, the sliding window mean method was used to resample the data, effectively eliminating the problems caused by inconsistent sampling frequencies of different sensors. Third, a Kalman filter was applied for data denoising and normalization, improving data quality. Regarding feature extraction, by selecting the most representative features for different types of sensor data, such as the mean and standard deviation of temperature and humidity, fluctuations in gas content, and energy characteristics of electromagnetic data, the time complexity was effectively reduced. According to the dimension, the computational efficiency is improved; in the multi-source data fusion stage, by calculating the importance of each sensor feature and using the entropy weight method for weighted fusion, the weights of different sensor data can be adaptively adjusted, giving full play to the advantages of various sensors and avoiding the excessive influence of some sensor data anomalies on the overall judgment; this invention not only effectively integrates information from different types of sensors, but also further improves the accuracy and reliability of identifying latent faults in switchgear through feature selection and weight optimization; compared with traditional single data processing methods, this invention can more comprehensively mine the information hidden in the data, providing more reliable and comprehensive features for subsequent fault classification and prediction, thereby improving the identification ability and prediction accuracy of various latent faults.

[0104] Specifically, in step D3, fault classification is performed based on the weighted fused feature vectors to obtain the fault classification results, including the following steps D31-D32:

[0105] D31: Defines typical fault categories for switchgear.

[0106] D32: The unknown state is classified using the support vector machine model based on the weighted fused feature vectors to obtain the fault classification result.

[0107] It should be noted that in switchgear fault classification, a support vector machine model combined with multi-source sensor data (such as temperature, vibration, gas content, electromagnetic characteristics, etc.) can be used to achieve accurate fault diagnosis. Specifically, by defining typical fault categories for switchgear, the support vector machine model is used to classify unknown states based on sensor data. The designed polynomial kernel function can be specifically represented by the following formula:

[0108] K(x i ,x j )=(γx i ·x j +r) d ;

[0109] Where K(x) i ,x j ) represents the output of the kernel function; x i and x j γ is the input; d is the order of the polynomial, used to control the complexity of the curve; γ controls the degree to which the data is mapped to a higher dimension; r is the offset, used to adjust the mapping effect.

[0110] In one alternative implementation, to ensure the accuracy of fault classification, the support vector machine model can be further evaluated for accuracy, precision, or recall, and the parameters of the support vector machine model can be optimized.

[0111] For example, when using a support vector machine for fault classification or prediction in step D32, the training objective of the model is to learn a mapping function f. model The support vector machine model can be specifically represented by the following formula:

[0112] y(t)=f model (f weight (t));

[0113] Where y(t) is the predicted output at time t, the classification label is "normal" or "fault" and continuous values ​​(such as the probability of fault occurrence, the expected fault time, etc.), and the mapping function is f. model Optimize based on training data, learn how to predict or classify faults based on sensor input features.

[0114] In one alternative implementation, fault classification is performed based on the weighted fused feature vectors to obtain the fault classification results. Alternatively, an ensemble learning method can be used to combine the prediction results of multiple base classifiers (such as random forests and gradient boosting decision trees) and use strategies such as voting or weighted averaging to form the final decision, effectively improving classification accuracy and system robustness.

[0115] In another alternative implementation, fault classification is performed based on the weighted fused feature vectors to obtain the fault classification results. Alternatively, deep learning networks based on attention mechanisms, such as a hybrid structure of convolutional neural networks and long short-term memory networks, can be applied to automatically learn the importance weights of temporal features, thereby achieving accurate classification of different fault modes of switchgear. Furthermore, it can adaptively focus on key time periods and sensor features, adapt to changes in fault modes, solve the sample imbalance problem through transfer learning, and improve the ability to identify fault types.

[0116] It should be noted that this invention defines typical fault categories for switchgear and uses a support vector machine model combined with a multinomial kernel function for fault classification, achieving accurate identification and classification of latent faults in switchgear. Compared with traditional threshold judgment methods, this invention has stronger adaptability and learning ability, continuously improving classification accuracy with data accumulation and adapting to changes in different working environments and equipment states. Furthermore, this invention can not only identify known fault types but also discover novel fault modes that have never appeared before through anomaly detection, expanding its applicability. The application of the support vector machine model also solves the shortcomings of traditional methods in handling small samples and imbalanced data, enabling this invention to effectively learn from limited samples even with limited historical fault data, demonstrating high accuracy and robustness in practical applications. This invention not only improves the early identification capability of latent faults but also predicts fault development trends, providing decision support for preventative maintenance of equipment, thereby effectively reducing downtime losses and safety risks caused by sudden equipment failures.

[0117] In this embodiment of the invention, fault identification is performed based on real-time monitoring data and fault classification results, including the following steps E1-E2:

[0118] E1: Based on real-time monitoring data and fault classification results, the abnormal status of the switchgear is classified into different levels.

[0119] E2: Based on the abnormal state level, a threshold judgment is made, and the device area is marked and warned according to the threshold judgment result.

[0120] Specifically, in step E1, based on real-time monitoring data and fault classification results, the abnormal state of the switchgear is classified into different levels. The specific operation can be as follows:

[0121] By identifying faults based on real-time monitoring data and fault classification results, the abnormal state of the air switch cabinet is divided into five levels. When the system detects that the abnormal state of a certain device exceeds the normal threshold, the monitoring screen marks the area of ​​the device as a red warning and pops up a warning sign. At the same time, it provides a real-time data query function, which makes it convenient for maintenance personnel to view the historical fault records and operating status of the air switch cabinet, and automatically generates a maintenance report.

[0122] Based on the system status of the switchgear, data support and optimization suggestions are provided for maintenance plans, such as suggested maintenance cycles, component replacement times, or fault prevention measures, to improve equipment safety and service life.

[0123] In one alternative implementation, fault identification based on real-time monitoring data and fault classification results can also employ a multi-dimensional anomaly measurement method. This involves constructing the normal operating envelope of each parameter of the switchgear, calculating the Mahalanobis distance or kernel density estimate of the real-time data deviation from the normal region, and combining a multi-parameter comprehensive anomaly index and threshold mechanism to achieve accurate location and risk classification of different types of faults.

[0124] In another alternative implementation, fault identification based on real-time monitoring data and fault classification results can also be achieved by combining knowledge graphs and inference engines. This involves constructing a knowledge network from switchgear fault characteristics, equipment topology relationships, historical fault cases, and expert experience, using inference algorithms to analyze fault propagation paths and root causes, and integrating fuzzy logic to handle uncertain information. This provides interpretable fault location results and handling suggestions, supporting maintenance personnel in making intelligent decisions.

[0125] Furthermore, the data storage and fault detection module also includes communication and human-computer interaction functions, the specific operations of which can be:

[0126] Wi-Fi wireless communication technology is used for data transmission to ensure the real-time nature and reliability of fault monitoring data, thereby achieving the purpose of remote monitoring and data exchange.

[0127] The collected sensor data, fault diagnosis results, alarm logs, and maintenance records are stored in the database to provide data support for subsequent intelligent analysis and predictive maintenance.

[0128] The system uses a graphical interface to display the operating status of the air switch cabinet, ensuring easy data readability, supporting real-time monitoring and dynamic data updates, and providing fault alarms.

[0129] It should be noted that this invention classifies the abnormal state of switchgear into five levels and uses threshold judgment to generate warning marks. It also provides real-time data query and maintenance report generation functions, achieving intuitive display and efficient management of fault information. The fault level classification allows the invention to adopt different response strategies according to the severity of the fault, avoiding overreaction to minor anomalies and underreaction to serious faults. The real-time warning function notifies maintenance personnel immediately upon the occurrence of a fault, shortening the fault response time. The graphical interface design presents status information in an intuitive and easy-to-read manner, reducing the cognitive burden on operators and improving work efficiency. The historical data query function allows maintenance personnel to trace the equipment's operating history, analyze the patterns and trends of fault occurrence, and provide a basis for preventative maintenance. The automatically generated maintenance report not only records fault information but also includes recommended maintenance measures, providing professional guidance for maintenance personnel. Through the organic combination of these functions, this invention not only meets the needs of daily monitoring and fault early warning of switchgear but also provides data support for the full lifecycle management of equipment. Compared with traditional manual inspection methods, this invention achieves automation, intelligence, and visualization of switchgear status monitoring, improving monitoring efficiency and accuracy while reducing labor costs.

[0130] In summary, this invention, by deploying multiple sensors within gas pipelines to comprehensively capture switchgear operating status information, employs Wi-Fi wireless communication technology for real-time data transmission, improves data quality and information utilization through data preprocessing, feature extraction, and multi-source data fusion, achieves accurate fault classification based on support vector machines, and finally displays the results and provides decision support through a graphical interface. This addresses the problems of insufficient information dimensions, poor data quality, and low diagnostic accuracy in traditional switchgear monitoring systems, enabling the early detection of abnormal signals and preventing potential problems. In particular, the application of multi-sensor fusion technology and machine learning methods enhances the ability to identify various latent faults, avoiding missed or false diagnoses caused by single-parameter monitoring. Through multi-sensor data fusion and intelligent analysis, early identification and accurate diagnosis of latent faults in switchgear are achieved, improving the accuracy and timeliness of fault diagnosis, reducing equipment downtime and maintenance costs, extending the service life of switchgear equipment, and simultaneously enhancing the safety and reliability of power system operation.

[0131] Example 3, referring to Figure 3This is the third embodiment of the present invention, which provides a remote diagnosis method for latent faults in switchgear, comprising: acquiring multidimensional data of the switchgear's operating status using a multidimensional sensor sensing module; transmitting the multidimensional data to a data fusion analysis module using a data transmission module; preprocessing the multidimensional data using the data fusion analysis module to obtain preprocessed multidimensional data; determining fault classification results based on the preprocessed multidimensional data; and identifying faults based on real-time monitoring data and fault classification results using a data storage and fault detection module.

[0132] Example 4 is the fourth embodiment of the present invention, which differs from the previous three embodiments in that: Figure 4 As shown, if the function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0133] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0134] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0135] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0136] Example 5, referring to Tables 1 to 4, is the fifth embodiment of the present invention, which provides a remote diagnostic system for latent faults in switchgear. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0137] This embodiment uses a 35kV switchgear testing device to comprehensively verify the multi-dimensional sensor sensing module and data fusion analysis module of the present invention under a controlled environment. The experiment simulates the occurrence and development process of latent faults by setting up a controllable heat source and a gas release device inside the switchgear. The multi-dimensional sensor sensing module uses a particle density sensor, a temperature and humidity sensor, a gas content sensor, and an electromagnetic radiation sensor to be deployed sequentially in the gas pipeline. The gas pump drives the gas circulation sampling to achieve full-dimensional monitoring of the internal environment of the switchgear. Each experiment lasts for 2 hours, and the data is collected every 30 minutes to ensure the repeatability and statistical validity of the data. The temperature and humidity data, gas content, and electromagnetic radiation energy are shown in Tables 1, 2, and 3.

[0138] Table 1 Temperature and Humidity Data

[0139] time Temperature (°C) humidity(%) state 10:00 35 40 normal 12:00 37 42 normal 14:00 45 55 Fault 16:00 47 60 Fault

[0140] Table 2 Gas Content Table

[0141] time Content (ppm) state 10:00 50 normal 12:00 52 normal 14:00 300 Fault 16:00 350 Fault

[0142] Table 3 Electromagnetic Radiation Energy Table

[0143] Frequency band (kHz) Energy (dB) normal Energy (dB) fault 0-1 -60 -45 1-5 -50 -30 5-10 -40 -20

[0144] As shown in Table 1, when a fault occurs inside the switchgear, the temperature rises from the normal 35-37℃ to 45-47℃, and the relative humidity also increases from 40%-42% to 55%-60%. The calculated heat dissipation power increases by more than 135%, verifying that the temperature and humidity sensor can effectively capture the thermal environment changes caused by the fault. Secondly, the gas content monitoring results show that the gas concentration under fault conditions increases by 6-7 times compared to the normal conditions. The increase in characteristic gases such as CO and NO2 indicates that there is abnormal arc discharge or insulation material deterioration inside the switchgear, proving the high sensitivity of the gas content sensor. Electromagnetic radiation monitoring data shows that the electromagnetic energy in all frequency bands is enhanced under fault conditions, especially the 1-10kHz band, which increases by 20dB, indicating that the characteristic frequencies extracted by Fourier transform can effectively identify the electromagnetic anomaly modes inside the switchgear.

[0145] To verify the effectiveness of the data fusion analysis module of this invention, the above data was input into the data fusion algorithm. The feature weights of each sensor were calculated by the entropy weight method and weighted fusion was performed. Then, the support vector machine model based on the polynomial kernel function was used for fault classification. The comparison data of different models are shown in Table 4.

[0146] Table 4. Model Comparison Data Table

[0147] Model Accuracy (%) Accuracy (%) Recall rate (%) This invention 95.3 94.7 96.1 Decision Tree 89.2 88.3 90.1

[0148] As shown in Table 4, the classification model based on multidimensional data fusion in this invention significantly outperforms traditional methods in all performance indicators. Specifically, the accuracy is improved by 6.1% compared to decision trees, 3.8% compared to random forests, and 2.2% compared to neural networks; precision is improved by 6.4%, 3.9%, and 2.3%, respectively; and recall is improved by 6.0%, 3.8%, and 2.3%, respectively. More importantly, the training time of this invention is only 27% of that of neural networks, demonstrating excellent computational efficiency. By adjusting the feature weights of each sensor using the entropy weight method, the excessive influence of single sensor data anomalies on the overall judgment is effectively avoided. This achieves efficient fusion of multi-source heterogeneous data and accurate fault classification, providing reliable technical support for the early identification and preventative maintenance of latent faults in switchgear.

[0149] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A remote diagnostic system for latent faults in switchgear, characterized in that: include, Multi-dimensional sensor perception module, data transmission module, data fusion and analysis module, and data storage and fault detection module; The multi-dimensional sensor sensing module is used to acquire multi-dimensional data on the operating status of the switchgear. The data transmission module is used to transmit the multidimensional data to the data fusion and analysis module; The data fusion and analysis module is used to preprocess the multidimensional data to obtain preprocessed multidimensional data, and to determine the fault classification result based on the preprocessed multidimensional data. The data storage and fault detection module is used to identify faults based on real-time monitoring data and the fault classification results.

2. The remote diagnostic system for latent faults in switchgear as described in claim 1, characterized in that: Data preprocessing of the multidimensional data includes: The multidimensional data is subjected to time synchronization processing to obtain time-synchronized multidimensional data; The multidimensional data after time synchronization is resampled to obtain the resampled multidimensional data. The resampled multidimensional data is denoised and normalized, and outlier data is detected and imputed.

3. The remote diagnostic system for latent faults in switchgear as described in claim 2, characterized in that: The fault classification results are determined based on the preprocessed multidimensional data, including: Data feature extraction is performed on the preprocessed multidimensional data to determine the feature data; Based on the aforementioned feature data, multi-source data fusion is performed to determine the weighted fused feature vector; Fault classification is performed based on the weighted and fused feature vectors to obtain the fault classification results.

4. The remote diagnostic system for latent faults in switchgear as described in claim 3, characterized in that: The fault identification based on real-time monitoring data and the fault classification results includes: Based on real-time monitoring data and fault classification results, the abnormal status of the switchgear is classified into different levels. Threshold judgments are made based on the level of abnormal status, and the equipment area is marked and warned according to the threshold judgment results.

5. The remote diagnostic system for latent faults in switchgear as described in claim 4, characterized in that: Data feature extraction is performed on the preprocessed multidimensional data to determine the feature data, including: Local temporal features are extracted from the preprocessed multidimensional data to obtain temporal features, frequency domain features, and temporal features.

6. The remote diagnostic system for latent faults in switchgear as described in claim 5, characterized in that: Based on the aforementioned feature data, multi-source data fusion is performed to determine the weighted fused feature vector, including: Calculate the importance of the time-domain features, frequency-domain features, and time-series features; Based on the importance, the time-domain features, frequency-domain features, and time-series features are concatenated and weighted to obtain a weighted fused feature vector.

7. The remote diagnostic system for latent faults in switchgear as described in claim 6, characterized in that: The construction of the multidimensional sensor sensing module includes: Particle density sensor, temperature and humidity sensor, gas content sensor and electromagnetic radiation sensor are arranged in the gas pipeline according to the arrangement order. An air pump is placed inside a gas pipeline to draw in external gas for sampling, thereby sensing the internal conditions of the switchgear.

8. A method for remote diagnosis of latent faults in switchgear, using a remote diagnosis system for latent faults in switchgear as described in any one of claims 1 to 7, characterized in that: include, Multidimensional sensor sensing modules are used to acquire multidimensional data on the operating status of the switchgear. The data transmission module is used to transmit multidimensional data to the data fusion and analysis module; The data fusion and analysis module is used to preprocess the multidimensional data to obtain preprocessed multidimensional data, and the fault classification results are determined based on the preprocessed multidimensional data. Fault identification is performed based on real-time monitoring data and fault classification results through the data storage and fault detection module.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the remote diagnostic system for latent faults in switchgear as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the remote diagnostic system for latent faults in switchgear as described in any one of claims 1 to 7.

Images