Circuit breaker intelligent operation and maintenance method and system based on multi-source data fusion

Abstract

The invention discloses a circuit breaker intelligent operation and maintenance method and system based on multi-source data fusion. The method comprises the following steps: acquiring and converging electrical, mechanical, visual, chemical, thermal and time series multi-dimensional monitoring data of a circuit breaker; preprocessing the multi-dimensional monitoring data; extracting the data characteristic quantity of each dimension from the preprocessed multi-dimension monitoring data; performing fusion processing on the data feature quantity of each dimension to form a comprehensive feature set; quantifying the influence degree of each data characteristic quantity on the state of the circuit breaker, and performing joint reasoning based on the comprehensive characteristic set to form a state research and judgment conclusion; and based on a state research and judgment conclusion, a decision mechanism is dynamically generated by adopting a multi-target decision model, and intelligent operation and maintenance of the circuit breaker based on multi-source data fusion are realized. The intelligent level of operation and maintenance work of the transformer substation circuit breaker can be improved, and the power grid operation and maintenance efficiency and equipment reliability are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of power equipment operation and maintenance, and particularly relates to a circuit breaker intelligent operation and maintenance method and system based on multi-source data fusion. BACKGROUND

[0002] In the large and complex operation architecture of modern power systems, substation circuit breakers undoubtedly play a key role in ensuring the safe and stable operation of power systems, and their reliable operation is crucial for the normal operation of the entire power network. Once a circuit breaker fails, it is likely to trigger a chain reaction, leading to a large-scale power outage, which will have a serious impact on social production and people's daily life.

[0003] Traditional circuit breaker operation and maintenance methods are mainly based on periodic manual inspection and relatively simple online monitoring methods. Periodic manual inspection, as a long-term operation and maintenance strategy, can to some extent discover some problems, but as the scale of the power system continues to expand and the operating environment becomes increasingly complex, its limitations become increasingly apparent, mainly including the following two aspects:

[0004] On the one hand, the inspection cycle is usually pre-set and relatively fixed. This fixed-cycle inspection mode is difficult to capture sudden fault hazards in the operation of circuit breakers in real time. For example, some key components inside the circuit breaker, such as contacts, springs, etc., may experience abnormal wear or accelerated aging under long-term high-load operation or harsh environmental conditions. During the interval between two inspections, these potential faults may further deteriorate, developing from initial minor abnormalities to serious faults, ultimately causing the device to malfunction, directly affecting the continuity of power supply, and causing great inconvenience to power users.

[0005] On the other hand, the effectiveness of manual inspection is subject to various factors of the inspection personnel. The professional level of the inspection personnel varies, and their understanding and judgment of the device state differ, which may lead to different conclusions about the same device state by different inspection personnel. At the same time, the work attitude of the inspection personnel also affects the quality of the inspection, and if negligence or laxity occurs in the work, important signs of failure may be missed. In addition, the environmental conditions during inspection cannot be ignored, such as in harsh weather (heavy rain, sand, high temperature, etc.) or complex geographical environments, the inspection personnel may not be able to fully and carefully inspect the equipment, thereby increasing the risk of missed detection or misjudgment.

[0006] To solve the above problems, simple online monitoring methods can obtain some electrical parameters such as current, voltage and power in real time, and provide some device operation information for operation and maintenance personnel. However, this monitoring method has obvious limitations. As a complex electrical device, the operating state of the circuit breaker is affected by multiple factors. In addition to electrical parameters, changes in mechanical properties, visual appearance, chemical properties and thermal properties may be important signs of potential failure. For example, during frequent opening and closing operations, mechanical parts of the circuit breaker may loosen or wear due to mechanical stress. These mechanical failure risks cannot be detected in time by electrical parameter monitoring. In addition, changes in the chemical properties of the internal insulation medium of the circuit breaker, such as aging and decomposition of insulating oil, leakage of sulfur hexafluoride gas or generation of decomposition products, may indicate potential risks in the device, but simple electrical parameter monitoring cannot detect these chemical property changes. Furthermore, from the perspective of thermal properties, local overheating may indicate problems such as poor contact inside the device, but simple online monitoring cannot comprehensively and accurately monitor the temperature of each part of the device.

[0007] More importantly, the data obtained by these simple monitoring methods are often in isolation and lack effective fusion analysis. The various monitoring data cannot be organically linked to form a comprehensive and systematic view of the device operating state. This makes it difficult for operation and maintenance personnel to accurately assess the health of the circuit breaker as a whole, and thus cannot provide a comprehensive and reliable basis for operation and maintenance decisions. For example, a single current data anomaly may be caused by multiple reasons, and it is difficult to determine the specific fault cause and location based on current data alone. If electrical parameters and mechanical performance data, chemical property data, etc. are fused and analyzed, the root cause of the fault can be more accurately determined, and more targeted operation and maintenance measures can be taken.

[0008] In summary, the existing circuit breaker operation and maintenance method has obvious shortcomings in terms of intelligence level, operation and maintenance efficiency and device reliability guarantee. With the rapid development of modern power systems towards intelligence, large-scale and complexity, higher requirements are put forward for circuit breaker operation and maintenance. The existing operation and maintenance method cannot meet the growing demand for intelligent and efficient operation and maintenance. Therefore, there is an urgent need for a new method and system that can improve the intelligence level of substation circuit breaker operation and maintenance, improve power grid operation and maintenance efficiency and device reliability, to ensure the safe, stable and efficient operation of the power system. SUMMARY

[0009] The purpose of the present application is to provide a circuit breaker intelligent operation and maintenance method and system based on multi-source data fusion, which can improve the intelligence level of substation circuit breaker operation and maintenance, improve power grid operation and maintenance efficiency and device reliability.

[0010] Technical scheme: the intelligent operation and maintenance method of the circuit breaker based on multi-source data fusion, comprising:

[0011] Collect and gather multi-dimensional monitoring data of circuit breaker electrical, mechanical, visual, chemical, thermal, time series;

[0012] Preprocess the multi-dimensional monitoring data; extract the data feature quantity of each dimension from the preprocessed multi-dimensional monitoring data; fuse the data feature quantity of each dimension to form a comprehensive feature set;

[0013] Quantify the influence degree of each data feature quantity on the state of the circuit breaker, and make a joint inference based on the comprehensive feature set to form a state research conclusion;

[0014] Based on the state research conclusion, a multi-objective decision model is used to dynamically generate a decision mechanism to realize intelligent operation and maintenance of the circuit breaker based on multi-source data fusion.

[0015] Further, the collection and gathering of multi-dimensional monitoring data of circuit breaker electrical, mechanical, visual, chemical, thermal, time series, comprising:

[0016] The electrical parameters are collected by high-precision current transformers and voltage transformers installed on the conductive loop of the circuit breaker, the mechanical state information is obtained by vibration sensors and displacement sensors arranged on the mechanical transmission components, the visual data is collected by multi-spectral cameras arranged outside the circuit breaker, the chemical data is collected by gas sensors arranged in the gas chamber inside the circuit breaker, and the thermal data is monitored by temperature sensors arranged at key heat generating parts of the circuit breaker;

[0017] The time synchronization of current transformers, voltage transformers, vibration sensors, displacement sensors, multi-spectral cameras, gas sensors and temperature sensors is realized by the time module to accurately record the operation time and fault time of the circuit breaker; a unified timestamp is stamped on the multi-dimensional data collected at each moment to form a time series data set associated with "data-time";

[0018] According to the unified timestamp, the various data in the time series data set are gathered, the data is transmitted to the edge computing node through wired or wireless communication mode, and the data is transmitted to the cloud server through optical fiber network for data classification storage.

[0019] Further, the preprocessing of the multi-dimensional monitoring data comprises:

[0020] The multi-dimensional monitoring data is subjected to format standardization processing, and data coding and units are unified; a combination cleaning technology based on statistical methods and machine learning algorithms is used to remove noise, abnormal values and repeated data in the multi-dimensional monitoring data after format standardization processing, so as to realize data cleaning;

[0021] Different types of data in the cleaned multi-dimensional monitoring data are subjected to normalization processing, the dimensional influence between data is eliminated, the dimensional data are made comparable, and a suitable normalization method is selected according to the distribution characteristics of different data;

[0022] After the multi-dimensional monitoring data is subjected to normalization processing, the missing values in the data acquisition process are filled according to the time sequence characteristics of the data and the correlation between the dimensional data by using a classification interpolation strategy, so as to ensure the continuity and integrity of the data.

[0023] Further, the data feature quantity of each dimension is extracted from the preprocessed multi-dimensional monitoring data, including:

[0024] Based on the time domain and frequency domain characteristics of electrical data, a multi-algorithm fusion electrical data feature extraction strategy is used to extract key feature quantities reflecting the equipment operating state from the electrical data in the preprocessed multi-dimensional monitoring data;

[0025] Based on the dynamic characteristics of mechanical data, mechanical feature quantities are extracted from the preprocessed multi-dimensional monitoring data in the mechanical state information, from the split and close operation process data and the steady state operation process data in the mechanical state information;

[0026] Deep learning algorithm is used to extract bottom layer features and high layer semantic features from the visual data in the preprocessed multi-dimensional monitoring data;

[0027] Chemical feature quantities are extracted from the SF6 gas data and insulating oil data in the chemical data in the preprocessed multi-dimensional monitoring data;

[0028] Data feature quantities reflecting the thermal state of the equipment are extracted from the temperature time series data and thermal imaging images in the thermal data in the preprocessed multi-dimensional monitoring data.

[0029] Further, the data feature quantity of each dimension is subjected to fusion processing to form a comprehensive feature set, including:

[0030] A dimension reduction algorithm is used to reduce the dimension of the high-dimensional single-dimensional feature set in the data feature quantity of each dimension, remove the redundancy and correlation between features, and reduce the data dimension;

[0031] Based on the importance of the data feature quantity of each dimension to the circuit breaker state evaluation, weights are assigned to the data feature quantity of each dimension after dimension reduction processing, and a comprehensive feature set fused with multi-dimensional information is formed by weighted summation; a feature effectiveness verification mechanism is introduced during the fusion process, and data feature quantities with a contribution degree to the state evaluation higher than a preset threshold are selected by variance analysis method to ensure that the comprehensive feature set is minimized and the most representative information is retained.

[0032] Further, the influence degree of each data feature quantity on the circuit breaker state is quantified, and joint reasoning is performed based on the comprehensive feature set to form a state research conclusion, including:

[0033] A feature quantity influence degree quantification weight determination mechanism combining objective data and subjective expert experience is established, and a multi-index evaluation method is used to combine massive historical fault data and domain expert experience to quantify the influence degree of each data feature quantity on the circuit breaker state, thereby obtaining the feature quantity weight;

[0034] Based on the comprehensive feature set and the feature quantity weight, a joint reasoning model combining a machine learning algorithm and an expert system is constructed; and the joint reasoning model is used to perform multi-dimensional reasoning on the circuit breaker state to obtain a joint reasoning result;

[0035] By accessing new operation data and fault cases in real time, model parameters and reasoning rules are continuously updated to continuously optimize the accuracy and reliability of the model;

[0036] According to the joint reasoning result, a structured circuit breaker state research conclusion is generated in combination with preset judgment thresholds and standards for different state levels.

[0037] Further, the joint reasoning model is used to perform multi-dimensional reasoning on the circuit breaker state to obtain a joint reasoning result, including:

[0038] The machine learning algorithm in the joint reasoning model is trained by historical fault data and normal operation data, which outputs a preliminary judgment result and a confidence level of the circuit breaker state according to the input comprehensive feature set and feature quantity weight; the expert system integrates the fault diagnosis experience of domain experts to establish a reasoning rule base covering the corresponding relationship between different feature combinations and fault types; the preliminary judgment result and the confidence level are verified and corrected through the reasoning rule base to finally judge the current state of the circuit breaker, including normal operation, potential fault hidden danger and fault state, and give the corresponding confidence or probability value, and the current state of the circuit breaker and the corresponding confidence or probability value are jointly used as the joint reasoning result.

[0039] Further, based on the state research conclusion, a multi-objective decision model is used to dynamically generate a decision mechanism to realize intelligent operation and maintenance of the circuit breaker based on multi-source data fusion, including:

[0040] A multi-objective decision-making model covering device reliability, operation and maintenance cost, power outage time, and safety risk is constructed.

[0041] The device reliability, operation and maintenance cost, power outage time, and safety risk targets are converted into unified mathematical expressions by using the target programming method and the utility function method; the priority and weight of each target are determined in combination with the power grid operation requirements and operation and maintenance management specifications;

[0042] The state research and judgment conclusion of the circuit breaker, real-time power grid operation parameters, operation and maintenance resource information, and device importance level are input into the multi-objective decision-making model for solving; in the solving process, an intelligent optimization algorithm is used, the number of iterations and the fitness function are set, and the optimal or suboptimal operation and maintenance strategy meeting the multi-objective constraints is searched, and the multi-objective decision-making model outputs the optimal or suboptimal operation and maintenance strategy.

[0043] A decision mechanism is generated according to the optimal or suboptimal operation and maintenance strategy, and accurate decision-making for the current state of the circuit breaker is realized.

[0044] The decision mechanism is sent to the operation and maintenance management system and the relevant executing department.

[0045] Further, the device reliability target is quantified by a fault occurrence rate reduction rate, the operation and maintenance cost target is quantified by the sum of repair costs and spare parts costs, the power outage time target is quantified by the planned power outage time and the fault power outage time, and the safety risk target is quantified by the operation and maintenance operation accident occurrence rate.

[0046] Based on the same inventive concept, one of the circuit breaker intelligent operation and maintenance systems based on multi-source data fusion of the present application comprises:

[0047] A multi-dimensional data acquisition module is used to acquire and converge multi-dimensional monitoring data of the circuit breaker in the electrical, mechanical, visual, chemical, thermal, and time sequence dimensions.

[0048] A data fusion module is used to pre-process the multi-dimensional monitoring data, extract data feature quantities of each dimension from the pre-processed multi-dimensional monitoring data, and fuse the data feature quantities of each dimension to form a comprehensive feature set.

[0049] A joint inference module is used to quantify the influence degree of each data feature quantity on the state of the circuit breaker, perform joint inference based on the comprehensive feature set, and form a state research and judgment conclusion.

[0050] An intelligent operation and maintenance module is used to generate a decision mechanism dynamically based on the state research and judgment conclusion by using a multi-objective decision-making model, and realize intelligent operation and maintenance of the circuit breaker based on multi-source data fusion.

[0051] Advantages: Compared with the prior art, the present application has the following significant technical effects:

[0052] (1) In the feature extraction and fusion process, with the help of large and small model combination, the complex relationship between data can be deeply mined, so that the system understands the status of the circuit breaker closer to the level of human experts, and is no longer limited to simple surface analysis of data, truly realizing intelligent operation and maintenance decision; (2) In the state research and joint reasoning link, using large models and artificial intelligence algorithms, the influence of each feature on the status of the circuit breaker can be automatically quantified, which is more objective and accurate than traditional artificial experience judgment; (3) Using graph neural networks and feature selection algorithms based on attention mechanisms, not only complex data structures can be processed, but also feature weight can be dynamically adjusted, so that the status of the circuit breaker can be accurately judged, providing a reliable basis for intelligent operation and maintenance, improving the efficiency of power grid operation and maintenance, and greatly reducing unnecessary manual inspection workload; (4) In traditional operation and maintenance, manual inspection requires a lot of manpower and material resources, and due to the limitation of inspection cycle, it is difficult to discover potential faults in time; the present application can accurately predict faults in advance through real-time multi-source data monitoring and intelligent analysis, and arrange operation and maintenance work targetedly, thereby reducing unnecessary human investment; (5) According to the operation and maintenance strategy dynamically generated by the intelligent decision model, the repair time and content can be accurately determined, over-repair or under-repair can be avoided, and the utilization efficiency of operation and maintenance resources can be improved; (6) When facing the change of power grid operation state, the dynamic decision mechanism can quickly respond. When the power grid load suddenly changes, the system can quickly adjust the operation and maintenance strategy according to the status of the circuit breaker and the demand of the power grid, such as increasing the monitoring frequency or arranging the repair in advance, to ensure the stable operation of the power grid and avoid power failure accidents caused by equipment failure, thereby effectively improving the overall operation and maintenance efficiency of the power grid; (7) The comprehensive multi-source data acquisition of the present application covers multiple dimensions such as electrical, mechanical, visual, chemical and thermal of the circuit breaker, which can timely capture the subtle changes of the equipment in all aspects and discover potential fault hidden dangers in advance. Through monitoring of the chemical data of the insulating gas, the early warning can be given in the early stage of the decline of the insulating performance, which is much earlier than the traditional way of detecting the fault or obvious electrical parameter abnormality, thereby greatly advancing the fault discovery time, gaining more time for equipment maintenance and reducing the risk of equipment sudden failure; (8) In the operation and maintenance strategy making, based on the multi-objective decision model and large model thinking chain, the factors such as equipment status, power grid demand and operation and maintenance cost are fully considered, and the operation and maintenance strategy made is more scientific and reasonable. Not only the reliable operation of the equipment can be ensured, but also the service life of the equipment can be prolonged through optimization of the operation and maintenance plan. According to the equipment health index and operating conditions, the component replacement time can be accurately arranged to avoid equipment failure caused by component aging or overuse, thereby significantly enhancing the reliability of the circuit breaker equipment and ensuring the stable operation of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 A flowchart of a circuit breaker intelligent operation and maintenance method based on multi-source data fusion disclosed in the embodiments of the present application is shown;

[0054] Figure 2 This is a flowchart of a state assessment method disclosed in an embodiment of the present invention;

[0055] Figure 3 This is a flowchart of an intelligent decision-making process disclosed in an embodiment of the present invention;

[0056] Figure 4 This is a schematic diagram of the structure of a circuit breaker intelligent operation and maintenance system based on multi-source data fusion disclosed in an embodiment of the present invention. Detailed Implementation

[0057] The technical solution of the present invention will now be described in detail with reference to specific embodiments and accompanying drawings.

[0058] Example 1

[0059] like Figure 1 As shown, the present invention provides an intelligent operation and maintenance method for circuit breakers based on multi-source data fusion, comprising the following steps:

[0060] S1. Multi-source data acquisition and aggregation: Collect and aggregate multi-dimensional monitoring data of circuit breakers, including electrical, mechanical, visual, chemical, thermal, and time series data.

[0061] The specific implementation process of step S1 is as follows:

[0062] S1.1 Electrical parameters are collected by high-precision current transformers and voltage transformers installed on the conductive circuit of the circuit breaker; mechanical status information is obtained by vibration sensors and displacement sensors installed on the mechanical transmission components; visual data is collected by multispectral cameras deployed outside the circuit breaker; chemical data is collected by gas sensors installed in the gas chamber inside the circuit breaker; and thermal data is monitored by temperature sensors placed in key heat-generating parts of the circuit breaker.

[0063] In this embodiment, high-precision Rogowski coil current transformers and capacitive voltage transformers are installed at key positions of the circuit breaker's conductive loop, such as near the contacts and bus connections, to accurately capture real-time changes in electrical parameters such as current and voltage. At key parts of the mechanical transmission mechanism, such as the opening and closing springs and connecting rods, piezoelectric vibration sensors and linear variable differential transformer displacement sensors are respectively installed to monitor the vibration and displacement changes of mechanical components during the opening and closing process. Outside the circuit breaker cabinet, high-definition industrial cameras are arranged to ensure that the appearance of the circuit breaker, including the integrity of the insulator and whether there are discharge marks, can be clearly captured. By installing gas sensors based on infrared absorption spectroscopy technology in the internal gas chamber of the circuit breaker, the chemical composition and concentration changes of insulating gases such as sulfur hexafluoride decomposition products can be monitored in real time. Meanwhile, fiber Bragg grating temperature sensors are evenly arranged at heat-prone locations such as contacts and connection points to accurately collect thermal data. In addition, to record the change pattern of data over time, high-precision data collectors are used to sample sensor data at set time intervals (e.g., once per second), thereby forming continuous time series data.

[0064] S1.2, through the time synchronization module, the current transformer, voltage transformer, vibration sensor, displacement sensor, multi-spectral camera, gas sensor, temperature sensor obtain the time synchronization of corresponding data, thereby accurately recording the operation time, fault time of the circuit breaker; for each time collected multi-dimensional data, a unified time stamp is stamped, forming a "data-time" associated time series data set.

[0065] In this embodiment, a separate time synchronization module is configured to achieve time synchronization with each collection unit through GPS / Beidou dual-mode time service, accurately recording the opening and closing operation time, operation frequency, fault occurrence time, fault duration, equipment operation time, maintenance record time, and other time-related data of the circuit breaker. At the same time, a unified time stamp is stamped for each time collected multi-dimensional data, forming a "data-time" associated time series data set, providing accurate information in the time dimension for subsequent analysis of device operation rules, fault development trends, and time sequence correlation of multi-source data.

[0066] S1.3, according to the unified time stamp, the multiple data in the time series data set are aggregated, the data is transmitted to the edge computing node through wired or wireless communication mode, and the data is transmitted to the cloud server through the optical fiber network for data classification storage.

[0067] In this embodiment, multiple data are aggregated according to a unified timestamp, and the data are transmitted to an edge computing node through wired or wireless communication, the edge computing node removes obviously erroneous or repeated data, and then transmits the processed data to a cloud server through a fiber network or a 5G network for data classification and storage.

[0068] S2, multi-source data analysis and feature fusion: preprocessing the multi-dimensional monitoring data; extracting the data feature quantity of each dimension from the preprocessed multi-dimensional monitoring data; fusing the data feature quantity of each dimension to form a comprehensive feature set.

[0069] The specific implementation process of step S2 is as follows:

[0070] S2.1, preprocessing the multi-dimensional monitoring data, the specific steps are as follows:

[0071] S2.1.1, first, the multi-dimensional monitoring data is standardized in format, and the data coding and units are unified; then, a combined cleaning technology based on statistical methods and machine learning algorithms is used to remove noise, outliers and repeated data in the multi-dimensional monitoring data after format standardization, and data cleaning is realized.

[0072] In this embodiment, a full-process data cleaning mechanism of "preprocessing-detection-removal-correction" is constructed. First, the multi-dimensional monitoring data is standardized in format, and the data coding and units are unified; then, a combined cleaning technology based on statistical methods and machine learning algorithms is used to remove noise, outliers and repeated data in the collected data; for numerical data, the abnormal data points obviously deviating from the normal range are identified and removed according to the criterion; for time series data, the sliding window method combined with trend analysis is used to identify mutation anomalies; for repeated data, the data fingerprint comparison technology is used for deduplication; for noise data caused by temporary sensor failure, the wavelet threshold denoising algorithm is used for filtering to ensure the authenticity of the cleaned data.

[0073] S2.1.2, normalize the different types of data in the cleaned multi-dimensional monitoring data, eliminate the dimensional influence between the data, make the dimensional data comparable, and select the appropriate normalization method according to the distribution characteristics of different data.

[0074] In this embodiment, for electrical parameters such as current and voltage, and mechanical parameters such as stroke and speed, which are uniformly distributed, the Min-Max normalization method is used to map them to the [0, 1] interval; for chemical parameters such as gas concentration and acid value, and thermal parameters such as temperature and temperature change rate, which are normally distributed, the Z-Score normalization method is used to convert them to standard normal distribution data with a mean of 0 and a standard deviation of 1, ensuring the accuracy of subsequent feature extraction.​

[0075] S2.1.3, after the multi-dimensional monitoring data is normalized, in view of the missing value problem in the data acquisition process, according to the time sequence characteristics of the data and the correlation between the dimensions of the data, a classification interpolation strategy is used to fill in the missing values, so as to ensure the continuity and integrity of the data.

[0076] In this embodiment, for discrete missing values with a missing rate lower than 5%, linear interpolation or spline interpolation method is used to fill in the missing values by using the change trend of adjacent data points; for data with a missing rate of 5%-20% and obvious time sequence characteristics, a machine learning interpolation algorithm based on long short-term memory network (LSTM) is used to accurately predict and fill in the missing values by learning the time sequence rules of historical data; for data with a missing rate higher than 20%, a data reacquisition mechanism is triggered and reference data of the same type of equipment is used for supplement, so as to ensure the reliability of the interpolated data.

[0077] S2.2, the data characteristic quantities of each dimension are extracted from the preprocessed multi-dimensional monitoring data, and the specific steps are as follows:

[0078] S2.2.1, based on the time domain and frequency domain characteristics of electrical data, a multi-algorithm fusion electrical data feature extraction strategy is used to extract key feature quantities reflecting the running state of the equipment from the electrical data in the preprocessed multi-dimensional monitoring data.

[0079] In this embodiment, for electrical data, the current peak value, voltage effective value, peak value coefficient, waveform distortion rate and other characteristics are extracted in the time domain, and the harmonic content, total harmonic distortion (THD) and other characteristics are extracted in the frequency domain through fast Fourier transform (FFT); for transient signals such as short-circuit current and operating overvoltage, the time-frequency entropy and energy entropy of the signals are extracted through wavelet transform.

[0080] S2.2.2, based on the dynamic characteristics of mechanical data, the mechanical characteristic quantities are extracted from the preprocessed multi-dimensional monitoring data of mechanical state information.

[0081] In this embodiment, for the opening and closing process data, the opening and closing time difference, contact bounce times, bounce duration, slope of the opening and closing speed curve and other characteristics are extracted; for the steady-state running data, the operating mechanism vibration frequency, vibration amplitude, peak factor and kurtosis of the vibration signal and other characteristics are extracted; for the contact pressure data, the pressure mean value, pressure fluctuation range, pressure change rate and other characteristics are extracted.

[0082] S2.2.3, using convolutional neural network (CNN) and other deep learning algorithms, the bottom layer features and high layer semantic features are extracted from the visual data in the preprocessed multi-dimensional monitoring data.

[0083] In this embodiment, the bottom layer features include the gray histogram features, texture features, and edge features of the image; the high layer semantic features are extracted through deep convolution layers and full connection layers of a convolutional neural network (CNN), and include the contact wear area ratio, contact gap value, area and shape features of discharge traces, insulator surface contamination degree features, and the like, so as to realize fine visual analysis of the appearance and internal state of the equipment.

[0084] S2.2.4, for chemical data in the preprocessed multi-dimensional monitoring data, extracting chemical characteristic quantities from SF6 gas data and insulating oil data in the chemical data.

[0085] In this embodiment, combined with the correlation between chemical indicators and insulation performance and internal faults, the following features are extracted from the SF6 gas data: SF6 purity, absolute content of each decomposition product (SO2, H2S, CO, etc.), ratio relationship between decomposition products (such as SO2 / H2S ratio), and gas concentration change rate; and the following features are extracted from the insulating oil data: acid value, moisture content, dielectric loss value, dissolved gas component content in oil, and gas generation rate.

[0086] S2.2.5, for thermal data in the preprocessed multi-dimensional monitoring data, extracting data characteristic quantities capable of reflecting the thermal state of the equipment from temperature time series data and thermal imaging images in the thermal data.

[0087] In this embodiment, based on thermal distribution data and temperature change law, data characteristic quantities capable of reflecting the thermal state of the equipment are extracted from the temperature time series data and the thermal imaging images, specifically as follows: extracting the maximum hot spot temperature, average temperature, temperature change rate, and temperature fluctuation amplitude from the temperature time series data; extracting the hot spot area, temperature difference between the hot spot and the surrounding environment, and thermal distribution uniformity coefficient from the thermal imaging images; and combining the thermal threshold value during normal operation of the equipment, evaluating the thermal stability and overheating fault risk of the equipment.

[0088] S2.3, fusing the data characteristic quantities of each dimension to form a comprehensive feature set, specifically including the following steps:

[0089] S2.3.1, using a dimension reduction algorithm to perform dimension reduction processing on the high-dimensional data characteristic quantities in the data characteristic quantities of each dimension, removing the redundancy and correlation between the features, and reducing the data dimension;

[0090] S2.3.2, based on the importance of the data feature quantity of each dimension to the circuit breaker state evaluation, weights are assigned to the data feature quantity of each dimension after dimension reduction processing, and a comprehensive feature set integrating multi-dimensional information is formed by weighted summation; a feature effectiveness verification mechanism is introduced during the fusion process, and data feature quantities with a contribution degree to state evaluation higher than a preset threshold are selected by variance analysis method to ensure that the comprehensive feature set is minimized and the most representative information is retained.

[0091] In this embodiment, a two-step fusion strategy of “dimension reduction and redundancy removal + weighted fusion” is adopted to fuse the extracted multi-dimensional feature quantities; first, dimension reduction algorithms such as principal component analysis (PCA) and linear discriminant analysis (LDA) are used to reduce the dimension of the high-dimensional single-dimensional feature set, remove the redundancy and correlation between features, and reduce the data dimension; then, based on the importance of each dimension feature to the circuit breaker state evaluation, weights are assigned to the features after dimension reduction, and a comprehensive feature set containing multi-dimensional information is formed by weighted summation; a feature effectiveness verification mechanism is introduced during the fusion process, and features with a contribution degree to state evaluation higher than a threshold are selected by ANOVA to ensure that the comprehensive feature set is both concise and retains the most representative information for circuit breaker state evaluation, thereby improving the efficiency and accuracy of subsequent analysis and decision-making.

[0092] S3, circuit breaker state research and joint reasoning: quantifying the influence degree of each data feature quantity on the circuit breaker state, joint reasoning based on the comprehensive feature set to form the state research conclusion.

[0093] As shown in Figure 2 , the specific implementation process of step S3 is as follows:

[0094] S3.1, establish a feature quantity influence degree quantification weight determination mechanism combining objective data and subjective expert experience, use multi-index evaluation methods such as analytic hierarchy process (AHP) and fuzzy comprehensive evaluation method, combine massive historical fault data and domain expert experience, and quantify the influence degree of each data feature quantity on the circuit breaker state to obtain the feature quantity weight; specifically as follows:

[0095] First, statistical analysis of historical fault data is performed to obtain the correlation degree of each data feature quantity and fault type as the basis for objective weight; then, a judgment matrix is constructed according to expert experience to obtain subjective weight; finally, the weighted average method is used to fuse the objective weight and subjective weight to form the final feature quantity weight.

[0096] S3.2, based on the comprehensive feature set and the feature quantity weight, a joint reasoning model of “machine learning algorithm + expert system” is constructed; the joint reasoning model is used for multi-dimensional reasoning of the circuit breaker state to obtain the joint reasoning result.

[0097] In this embodiment, the circuit breaker state is inferred in multiple dimensions using a joint inference model to obtain a joint inference result, as follows:

[0098] The machine learning algorithm in the joint inference model is trained using historical fault data and normal operation data. The algorithm outputs a preliminary judgment result and a confidence level of the circuit breaker state based on the input set of comprehensive features and feature weight.

[0099] The expert system integrates the fault diagnosis experience of domain experts to establish a reasoning rule base containing hundreds of rules. The reasoning rule base covers the corresponding relationship between different feature combinations and fault types.

[0100] The preliminary judgment result and confidence level are verified and corrected through the reasoning rule base to ultimately determine the current state of the circuit breaker, including normal operation, potential fault risks (mild, moderate, and severe), and fault state, with the corresponding confidence level (accurate to 0.01) or probability value. The current state of the circuit breaker and the corresponding confidence level or probability value are jointly used as the joint inference result.

[0101] The machine learning algorithm uses intelligent algorithms such as Bayesian networks, neural networks, and support vector machines.

[0102] S3.3. Continuously update model parameters and reasoning rules by accessing new operation data and fault cases in real time to continuously optimize the accuracy and reliability of the model.

[0103] In this embodiment, the joint inference model has online learning capability. By accessing new operation data and fault cases in real time, the model parameters and reasoning rules are continuously updated to continuously optimize the accuracy and reliability of the model.

[0104] S3.4. According to the joint inference result, combine the preset judgment thresholds and standards for different state levels to generate a structured circuit breaker state research conclusion.

[0105] In this embodiment, the circuit breaker state research conclusion includes the overall operation state level of the device, whether there is an anomaly, the corresponding fault type of the anomaly, the specific location of the fault, the severity of the fault, and the fault development trend prediction. Each research conclusion is supported by the specific numerical value of the key abnormal feature, the weight, and the associated reasoning rule.

[0106] In this embodiment, a conclusion generation mechanism of "multi-level threshold + fault location" is established, and according to the joint reasoning result, the judgment threshold and standard of different state levels are combined to generate a structured circuit breaker state research conclusion. The conclusion content includes the overall operation state level of the device, whether there is an abnormality, the fault type corresponding to the abnormality, the specific location of the fault, the fault severity, and the fault development trend prediction. And for each research conclusion, support evidence is attached, that is, the specific value, weight and associated reasoning rule of the key abnormal characteristic quantity, to ensure the traceability of the conclusion and provide accurate and comprehensive basis for subsequent operation and maintenance decision.

[0107] S4, intelligent decision of circuit breaker operation and maintenance strategy: based on the state research conclusion, a multi-objective decision model is used to dynamically generate a decision mechanism to realize intelligent operation and maintenance of the circuit breaker based on multi-source data fusion.

[0108] As shown in Figure 3 , the specific implementation process of step S4 is as follows:

[0109] S4.1, a multi-objective decision model covering device reliability, operation and maintenance cost, outage time, and safety risk is constructed. The input of the model includes state research conclusion, power grid operation parameter, operation and maintenance resource information, and device importance level, and the output is the optimal operation and maintenance strategy.

[0110] S4.2, the objective programming method and the utility function method are used to convert the device reliability, operation and maintenance cost, outage time, and safety risk targets into unified mathematical expressions, and all indexes are standardized and the value range is [0, 1];

[0111] (1) Device reliability target, i.e. fault occurrence rate reduction rate :

[0112] ;

[0113] Wherein, represents the fault occurrence rate of the circuit breaker before operation (times / year), which is based on historical fault data statistics;

[0114] represents the expected fault occurrence rate (times / year) after operation, which is output by combining the state research conclusion such as insulation aging degree and mechanical characteristic parameter with the machine learning prediction model; The value range of , is closer to 1, the reliability improvement effect is more significant.

[0115] (2) Operation and maintenance cost target, standardized total operation and maintenance cost :

[0116] ;

[0117] wherein, is the total maintenance cost; is the repair cost, including manual repair cost, tool usage cost, outsourcing service cost, etc.; is the spare parts cost, including replacement component procurement cost, transportation cost, warehousing cost, etc.; is the personnel cost, including operation and maintenance personnel salary, training cost, etc.; is the maximum total cost of historical operation and maintenance of the same type circuit breaker, or an upper limit value set based on the operation and maintenance budget; the value range of is , the closer to 0, the better the cost control effect.

[0118] (3) Power outage time target, standardized total power outage time :

[0119] ;

[0120] wherein, is the planned maintenance power outage time, in h; is the expected failure power outage time, in h; , is the average single failure repair time, in h; is the maximum annual power outage time allowed by the device, in h, set according to the power grid dispatching specification, device importance level, etc. (such as critical circuit breaker ≤8h / year); the value range of is [0, 1], the closer to 0, the smaller the power outage impact.

[0121] (4) Safety risk target, standardized safety risk value :

[0122] ;

[0123] wherein, is the calculation of the operation and maintenance accident rate, , is the risk correction coefficient, which is adjusted based on the current operation environment (such as high temperature, high humidity), personnel qualifications (such as the rate of holding a certificate), equipment status (such as whether there are major defects) , the higher the risk , the larger the value; is the maximum historical accident rate of the same type of operation and maintenance, such as setting , i.e. 5%; the value range of is [0, 1], the closer to 0,

[0124] (5) Objective utility transformation function

[0125] ;

[0126] in, For the first The utility value of an objective. These are the standardized quantitative indicators mentioned above.

[0127] (6) Total Deviation Objective Function

[0128] ;

[0129] in, > > > Based on the objectives and their priorities, and in accordance with power grid operation requirements and maintenance management standards, the priorities and weights of each objective are determined, and they are ranked according to power operation and maintenance priorities, including safety risks. Highest, followed by reliability Power outage time Operation and maintenance costs ; For the first A positive deviation of the target utility value from the ideal value A negative deviation below the ideal value; such as safety risk utility. If the ideal value is 0.95, This represents reality Deviations less than 0.95 should be minimized first.

[0130] (7) Constraints

[0131] ;

[0132] in, For the first The ideal utility value for each objective can be set according to industry standards, such as security risk. ,reliability The constraints limit the reasonable range of utility values, and by linking utility values ​​with ideal values ​​through deviation variables, the constraint logic of goal programming is continued.

[0133] Among them, the equipment reliability target is quantified by the failure rate reduction rate, the operation and maintenance cost target is quantified by the sum of maintenance costs, spare parts costs, etc., the power outage time target is quantified by the planned power outage duration and the power outage duration due to failure, and the safety risk target is quantified by the operation and maintenance accident rate.

[0134] In this embodiment, based on the core demand of circuit breaker operation and maintenance, a multi-objective decision model covering multiple conflicting objectives such as device reliability, operation and maintenance cost, power outage time, safety risk, etc. is constructed; the input of the model includes state research and judgment conclusion, power grid operation parameters, operation and maintenance resource information, etc., and the output is the optimal operation and maintenance strategy; methods such as goal programming method and utility function method are used to convert multiple objectives into unified mathematical expressions, among which the device reliability objective is quantified by the fault occurrence rate reduction rate, the operation and maintenance cost objective is quantified by the sum of repair costs and spare parts costs, the power outage time objective is quantified by the planned power outage time and the fault power outage time, and the safety risk objective is quantified by the operation and maintenance operation accident rate; the priority and weight of each objective are determined in combination with the power grid operation requirements and operation and maintenance management specifications.

[0135] S4.3, a dynamic decision mechanism is established: the state research and judgment conclusion of the circuit breaker and the real-time power grid operation parameters, operation and maintenance resource information, device importance level are input into the multi-objective decision model for solving; in the solving process, intelligent optimization algorithms such as genetic algorithm and particle swarm optimization algorithm are used, and parameters such as the number of iterations and the fitness function of the intelligent optimization algorithm are set to search for the optimal or suboptimal operation and maintenance strategy that meets the multi-objective constraints. The multi-objective decision model outputs the optimal or suboptimal operation and maintenance strategy.

[0136] In this embodiment, a dynamic decision mechanism of "real-time data driving + intelligent algorithm solving" is established, based on the circuit breaker state research and judgment conclusion, combined with real-time factors such as real-time power grid operation, device importance level, and operation and maintenance resource availability, input into the multi-objective decision model for solving; genetic algorithm, particle swarm optimization algorithm and other intelligent optimization algorithms are used, and parameters such as the number of iterations and the fitness function are set to search for the optimal or suboptimal operation and maintenance strategy that meets the multi-objective constraints.

[0137] S4.4, generate a decision mechanism based on the optimal or suboptimal operation and maintenance strategy to achieve accurate decision-making for the current circuit breaker state.

[0138] The generated decision mechanism includes specific operation and maintenance content, for different state conditions, generate monitoring scheme, preventive maintenance plan, emergency repair scheme and other operation and maintenance content, as follows: for mild hidden danger, generate a regular monitoring scheme, and specify the monitoring frequency and key indicators; for moderate hidden danger, generate a preventive maintenance plan, and specify the maintenance time, steps and required resources; for fault state, generate an emergency repair scheme, and specify the repair process, personnel division of labor and spare parts allocation scheme, to achieve accurate decision-making for the current circuit breaker state.

[0139] S4.5, construct a closed-loop management process of decision-making and feedback optimization: send the decision mechanism to the operation and maintenance management system and the relevant execution department; the system analyzes the operation process data in real time, evaluates the implementation effect of the operation and maintenance strategy; according to the feedback information in the operation and maintenance practice, establish a model optimization mechanism, adjust and optimize the target weight, constraint condition and algorithm parameter of the multi-objective decision model, and update the expert experience library and fault case library, continuously improve the accuracy and adaptability of intelligent decision-making, in order to realize the continuous improvement of the intelligent operation and maintenance of the circuit breaker based on multi-source data fusion.

[0140] In this embodiment, a closed-loop management process of "decision-making-execution-feedback-optimization" is constructed, and the generated operation and maintenance strategy is sent to the operation and maintenance management system and the relevant execution department through a standardized interface; the system analyzes the operation process data in real time, evaluates the implementation effect of the operation and maintenance strategy, such as fault repair rate, equipment operation stability after maintenance, operation and maintenance cost control, etc.; and according to the feedback information in the operation and maintenance practice, a model optimization mechanism is established, the target weight, constraint condition and algorithm parameter of the multi-objective decision model are adjusted and optimized, and the expert experience library and fault case library are updated, the accuracy and adaptability of intelligent decision-making are continuously improved, in order to realize the continuous improvement of the intelligent operation and maintenance of the circuit breaker based on multi-source data fusion.

[0141] The present application realizes comprehensive and intelligent acquisition and processing of circuit breaker operation state data by combining multi-source data acquisition with artificial intelligence technology. From traditional dependence on single data type and simple manual inspection, it is changed to automatic acquisition of all-dimensional multi-dimensional data, and intelligent analysis is carried out by using deep learning algorithm, which greatly improves the intelligent degree of circuit breaker operation and maintenance.

[0142] Embodiment 2

[0143] As shown in Figure 4 , a circuit breaker intelligent operation and maintenance system based on multi-source data fusion of the present application, characterized in that it comprises:

[0144] A multi-dimensional data acquisition module is used for acquiring and converging multi-dimensional monitoring data of circuit breaker electrical, mechanical, visual, chemical, thermal and time series;

[0145] A data fusion module is used for pre-processing the multi-dimensional monitoring data; extracting data feature quantities of each dimension from the pre-processed multi-dimensional monitoring data; and fusing and processing the data feature quantities of each dimension to form a comprehensive feature set;

[0146] A joint inference module is used for quantifying the influence degree of each data feature quantity on the state of the circuit breaker, performing joint inference based on the comprehensive feature set, and forming a state judgment conclusion;

[0147] The intelligent operation and maintenance module is used for generating a decision mechanism dynamically based on a multi-target decision model and a state research and judgment conclusion, so as to realize intelligent operation and maintenance of the circuit breaker based on multi-source data fusion.

[0148] In an optional embodiment, the intelligent operation and maintenance method of the circuit breaker based on multi-source data fusion comprises: a) collecting and gathering multi-dimensional monitoring data of the circuit breaker in electrical, mechanical, visual, chemical, thermal and time sequence dimensions; b) pre-processing the multi-dimensional monitoring data and extracting data characteristic quantities of each dimension; forming a comprehensive feature set through fusion processing; c) quantifying the influence degree of each data characteristic quantity on the state of the circuit breaker, jointly reasoning based on the comprehensive feature set to form a state research and judgment conclusion; and d) generating a decision mechanism dynamically based on the state research and judgment conclusion by using a multi-target decision model, so as to realize intelligent operation and maintenance of the circuit breaker based on multi-source data fusion.

Claims

1. A method for intelligent operation and maintenance of circuit breakers based on multi-source data fusion, characterized in that, include: Collect and aggregate multi-dimensional monitoring data of circuit breakers, including electrical, mechanical, visual, chemical, thermal, and time series data. Preprocess the multi-dimensional monitoring data; Extract data features from each dimension of the preprocessed multi-dimensional monitoring data; The data features from various dimensions are fused and processed to form a comprehensive feature set; Quantify the impact of each data feature on the circuit breaker status, and perform joint reasoning based on the comprehensive feature set to form a status judgment conclusion; Based on the state assessment conclusions, a multi-objective decision-making model is adopted to dynamically generate a decision-making mechanism, thereby realizing intelligent operation and maintenance of circuit breakers based on multi-source data fusion.

2. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 1, characterized in that, The collection and aggregation of multi-dimensional monitoring data on circuit breakers, including electrical, mechanical, visual, chemical, thermal, and time-series data, includes: Electrical parameters are collected by high-precision current transformers and voltage transformers installed on the conductive circuit of the circuit breaker; mechanical status information is obtained by vibration sensors and displacement sensors installed on the mechanical transmission components; visual data is collected by multispectral cameras deployed outside the circuit breaker; chemical data is collected by gas sensors installed in the gas chamber inside the circuit breaker; and thermal data is monitored by temperature sensors placed in key heat-generating parts of the circuit breaker. The timing module enables the synchronization of data acquired by current transformers, voltage transformers, vibration sensors, displacement sensors, multispectral cameras, gas sensors, and temperature sensors, thereby accurately recording the operation time and fault time of the circuit breaker; and assigns a unified timestamp to the multi-dimensional data collected at each moment, forming a time-series dataset with data-time correlation. According to a unified timestamp, various data in the time series dataset are aggregated, and the data is transmitted to edge computing nodes via wired or wireless communication, and then transmitted to cloud servers via fiber optic networks for data classification and storage.

3. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 1, characterized in that, The preprocessing of multi-dimensional monitoring data includes: The multi-dimensional monitoring data is format-standardized to unify data coding and units; a combined cleaning technique based on statistical methods and machine learning algorithms is used to remove noise, outliers and duplicate data from the format-standardized multi-dimensional monitoring data to achieve data cleaning. Normalization processing is performed on different types of data in the cleaned multidimensional monitoring data to eliminate the influence of dimensions between data, so that the data in each dimension are comparable, and appropriate normalization methods are selected according to the distribution characteristics of different data. After normalizing the multi-dimensional monitoring data, to address the issue of missing values ​​during data collection, a classification interpolation strategy is used to fill in the missing values ​​based on the time series characteristics of the data and the correlation between the data in each dimension, thus ensuring the continuity and integrity of the data.

4. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 1, characterized in that, The extraction of data feature quantities for each dimension from the preprocessed multi-dimensional monitoring data includes: Based on the time and frequency domain characteristics of electrical data, a multi-algorithm fusion electrical data feature extraction strategy is adopted to extract key feature quantities that can reflect the operating status of equipment from the electrical data in the preprocessed multi-dimensional monitoring data. Based on the dynamic characteristics of mechanical data, mechanical feature quantities are extracted from the opening and closing operation process data and steady-state operation process data in the pre-processed multi-dimensional monitoring data of mechanical state information. Using deep learning algorithms, low-level features and high-level semantic features are extracted from visual data in preprocessed multi-dimensional monitoring data. For chemical data in preprocessed multidimensional monitoring data, chemical characteristic quantities are extracted from SF6 gas data and insulating oil data. For the thermal data in the preprocessed multi-dimensional monitoring data, extract data features that can reflect the thermal state of the equipment from the temperature time series data and thermal imaging maps.

5. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 1, characterized in that, The process of fusing data features from various dimensions to form a comprehensive feature set includes: A dimensionality reduction algorithm is used to reduce the dimensionality of high-dimensional data features in each dimension, thereby removing redundant information and correlations between features and reducing the data dimensionality. Based on the importance of data features in each dimension to circuit breaker condition assessment, weights are assigned to the data features after dimensionality reduction, and a comprehensive feature set integrating multi-dimensional information is formed by weighted summation. A feature validity verification mechanism is introduced during the fusion process, and data features with a contribution to condition assessment higher than a preset threshold are selected by variance analysis to ensure that the comprehensive feature set is minimized and the most representative information is retained.

6. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 1, characterized in that, The degree of influence of each data feature on the circuit breaker state is quantified, and joint reasoning is performed based on the comprehensive feature set to form a state judgment conclusion, including: Establish a mechanism for determining the weight of the influence of feature quantities by combining objective data and subjective expert experience. Use a multi-index evaluation method, combined with massive historical fault data and domain expert experience, to quantify the influence of each data feature quantity on the circuit breaker status, thereby obtaining the feature quantity weight. Based on the comprehensive feature set and feature weights, a joint reasoning model combining machine learning algorithms and expert systems is constructed; the joint reasoning model is used to perform multi-dimensional reasoning on the circuit breaker state to obtain the joint reasoning result. By continuously accessing new operational data and failure cases in real time, the model parameters and inference rules are constantly updated to continuously optimize the accuracy and reliability of the model. Based on the joint reasoning results, and combined with the preset judgment thresholds and standards for different state levels, a structured conclusion on the state assessment of the circuit breaker is generated.

7. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 6, characterized in that, The method of using a joint reasoning model to perform multi-dimensional reasoning on the circuit breaker state to obtain joint reasoning results includes: The machine learning algorithm in the joint inference model is trained using historical fault data and normal operation data. Based on the input comprehensive feature set and feature weights, the algorithm outputs a preliminary judgment result and confidence level of the circuit breaker status. The expert system integrates the fault diagnosis experience of domain experts and establishes an inference rule base covering the correspondence between different feature combinations and fault types. The preliminary judgment result and confidence level are verified and corrected through the inference rule base, and finally the current status of the circuit breaker is judged, including normal operation, potential fault hazards, and fault status, and the corresponding confidence level or probability value is given. The current status of the circuit breaker and the corresponding confidence level or probability value are used together as the joint inference result.

8. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 1, characterized in that, The aforementioned intelligent operation and maintenance of circuit breakers based on multi-objective decision-making mechanism, using a multi-objective decision-making model to dynamically generate decisions based on multi-source data fusion, includes: Construct a multi-objective decision-making model that covers multiple conflicting objectives such as equipment reliability, operation and maintenance costs, power outage time, and safety risks; By employing goal programming and utility function methods, the objectives of equipment reliability, operation and maintenance costs, power outage time, and safety risks are transformed into unified mathematical expressions; and by combining power grid operation requirements and operation and maintenance management specifications, the priority and weight of each objective are determined. The circuit breaker status assessment conclusions, real-time power grid operation parameters, maintenance resource information, and equipment importance level are input into a multi-objective decision model for solution. During the solution process, an intelligent optimization algorithm is used, with the number of iterations and fitness function set to search for the optimal or suboptimal maintenance strategy that satisfies the multi-objective constraints. The multi-objective decision model outputs the optimal or suboptimal maintenance strategy. A decision-making mechanism is generated based on the optimal or suboptimal operation and maintenance strategy to achieve accurate decision-making for the current circuit breaker status; The decision-making mechanism is sent to the operation and maintenance management system and relevant implementation departments.

9. The intelligent operation and maintenance method for circuit breakers based on multi-source data fusion according to claim 8, characterized in that, The equipment reliability target is quantified by the failure rate reduction rate, the operation and maintenance cost target is quantified by the sum of maintenance costs and spare parts costs, the power outage time target is quantified by the planned power outage duration and the power outage duration due to failure, and the safety risk target is quantified by the operation and maintenance accident rate.

10. A circuit breaker intelligent operation and maintenance system based on multi-source data fusion, characterized in that, include: The multi-dimensional data acquisition module is used to collect and aggregate multi-dimensional monitoring data of circuit breakers, including electrical, mechanical, visual, chemical, thermal, and time series data. The data fusion module is used to preprocess multi-dimensional monitoring data; extract data features of each dimension from the preprocessed multi-dimensional monitoring data; and fuse the data features of each dimension to form a comprehensive feature set. The joint reasoning module is used to quantify the impact of each data feature on the circuit breaker state, and to perform joint reasoning based on the comprehensive feature set to form a state judgment conclusion. The intelligent operation and maintenance module is used to dynamically generate a decision-making mechanism based on the status assessment conclusions and a multi-objective decision-making model, so as to realize intelligent operation and maintenance of circuit breakers based on multi-source data fusion.

Images