Intelligent monitoring system for transformer windings based on cuprous sulfide deposition characteristics

Abstract

The present invention discloses an intelligent transformer winding monitoring system based on cuprous sulfide deposition characteristics, including: a monitoring setting unit, used to create an online monitoring platform related to cuprous sulfide deposition characteristics in a target area; an acquisition unit, used to acquire multi-physical parameter monitoring data of transformer windings based on the online monitoring platform, the monitoring data including electromagnetic parameters, ultrasonic characteristics, light radiation characteristics, temperature and pressure information; an analysis unit, used to perform multi-dimensional analysis on multi-physical parameter monitoring data based on a deep learning analysis model to determine potential discharge defects in transformer windings; a diagnosis unit, used to automatically trigger an early warning when a potential discharge defect is detected, and to collect all relevant monitoring data in the target area to form a comprehensive diagnosis report. The maintenance efficiency of the equipment is greatly improved, and significant economic losses caused by failure delays are avoided.

Description

Technical Field

[0001] The invention relates to the technical field of power equipment monitoring, and in particular to an intelligent transformer winding monitoring system based on cuprous sulfide deposition characteristics. Background Art

[0002] In recent years, safety accidents caused by transformer insulation failure due to corrosive sulfur in transformer oil have frequently occurred at home and abroad, seriously affecting the safe and reliable operation of the power system. The active sulfide in the transformer oil reacts with the copper wire to form cuprous sulfide (Cu 2 S), which is deposited on the copper wire and insulation paper surface between the winding turns, reducing the electrical performance of the insulation paper. As cuprous sulfide accumulates, the insulation performance further deteriorates, eventually causing inter-turn insulation discharge. This discharge can cause local insulation breakdown, generate high-energy arc faults, cause local pressure surges inside the transformer, and even lead to serious accidents such as oil tank rupture and explosion.

[0003] The current relay protection device is mainly for large fault current and high energy action, which belongs to passive protection. This protection method cannot sensitively detect the initial stage of the fault, cannot quickly remove the fault in the early stage, and cannot meet the detection needs of early faults such as inter-turn discharge. The existing online monitoring device lacks the ability to accurately distinguish serious defects that occur in a short time scale (days to hours), and it is difficult to timely detect the inter-turn discharge fault caused by cuprous sulfide deposition. The inter-turn fault has the characteristics of complex defect causes, fast development speed, and serious consequences. The existing technology is difficult to capture and diagnose such faults in time, resulting in the rapid deterioration of the fault without being detected. The internal structure of the transformer is complex, including windings, cores, outgoing wires, insulation and supports. The physical signals monitored externally not only depend on the location and intensity of the discharge, but also are affected by the attenuation and interference of the signal during the propagation process inside the transformer. The existing monitoring methods lack in-depth research on the propagation characteristics of these multi-physical signals, resulting in insufficient accuracy and reliability of fault diagnosis.

[0004] Therefore, there is an urgent need for an intelligent transformer winding monitoring system based on cuprous sulfide deposition characteristics. Summary of the invention

[0005] The present invention provides a transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics to solve the above problems existing in the prior art.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] Transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics, including:

[0008] A monitoring setting unit for creating an online monitoring platform related to the characteristics of cuprous sulfide deposition in the target area;

[0009] An acquisition unit is used to acquire multi-physical parameter monitoring data of transformer windings based on an online monitoring platform, the monitoring data including electromagnetic parameters, ultrasonic characteristics, optical radiation characteristics, temperature and pressure information;

[0010] An analysis unit, which is used to perform multi-dimensional analysis on multi-physical parameter monitoring data based on a deep learning analysis model to determine potential discharge defects in transformer windings;

[0011] The diagnostic unit is used to automatically trigger an early warning when a potential discharge defect is detected, and to collect all relevant monitoring data in the target area to form a comprehensive diagnostic report.

[0012] The monitoring setting unit includes:

[0013] The platform construction module is used to build a real-time online monitoring platform in the target area, integrate various monitoring modules, and form a monitoring network covering various areas and key parts of the transformer winding;

[0014] Module integration module for integrating different monitoring modules related to cuprous sulfide deposition characteristics into an online monitoring platform;

[0015] The monitoring control module is used to coordinate different monitoring modules, perform joint monitoring operations on the target area, and obtain real-time monitoring data of cuprous sulfide deposition.

[0016] The acquisition unit includes:

[0017] Electromagnetic parameter monitoring module, used to obtain electromagnetic parameters of transformer windings, including current fluctuations and voltage mutations, to identify discharge phenomena;

[0018] Ultrasonic characteristic monitoring module, which is used to monitor the local stress changes inside the winding using ultrasonic detection technology and identify physical changes caused by cuprous sulfide deposition;

[0019] An optical radiation characteristics monitoring module for detecting abnormal features in the spectrum using spectral analysis techniques, including changes in infrared and ultraviolet light, to identify signs of high temperature and partial discharge;

[0020] Temperature and pressure information monitoring module, used to measure transformer oil temperature and internal pressure, and analyze winding temperature rise and pressure changes caused by cuprous sulfide deposition;

[0021] The data synchronization module is used to synchronize the electromagnetic parameters, ultrasonic characteristics, optical radiation characteristics, temperature and pressure information to ensure the coordination and consistency of multi-physical parameter monitoring data.

[0022] Among them, potential discharge defects in transformer windings are determined, including:

[0023] Identify and obtain current transformer monitoring data from the multi-physical parameter monitoring data, wherein the current transformer monitoring data includes monitoring parameters and monitoring environment parameters;

[0024] According to the monitoring parameters and monitoring environment parameters, the current transformer monitoring data is processed in layers to obtain the monitoring data to be analyzed;

[0025] Obtaining transformer fault standard data, and obtaining fault characteristic data from the transformer fault standard data;

[0026] After noise elimination processing is performed on the fault feature data, a data feature vector is constructed, and fault-related data is obtained from the transformer fault standard data;

[0027] According to the fault-related data, the data feature vectors are clustered to obtain a data group to be trained, and the initial model is trained with the data group to obtain a deep learning analysis model;

[0028] Input the monitoring data to be analyzed into the preset deep learning analysis model one by one to obtain the transformer status benchmark, including:

[0029] Inputting the first monitoring data to be analyzed into the deep learning analysis model, performing similarity matching on the first monitoring data to be analyzed in the deep learning analysis model, and outputting a first analysis result according to the matching result;

[0030] Starting from the second monitoring data to be analyzed, the previous analysis result is input into the deep learning analysis model to obtain the corresponding analysis result, and the first analysis result and all analysis results are integrated to obtain the transformer status benchmark;

[0031] The transformer status benchmark and current transformer monitoring data are input into the deep learning analysis model to obtain the monitoring result data of potential discharge defects in the transformer winding.

[0032] Among them, a comprehensive diagnostic report is formed, including:

[0033] Set dynamic thresholds. When monitoring parameters exceed the preset normal range, the monitoring result data of potential discharge defects will automatically trigger an early warning.

[0034] After the early warning is triggered, all relevant monitoring data are automatically collected in the target area to form a data set to be analyzed;

[0035] The data set to be analyzed is input into the deep learning analysis model to automatically process and analyze the collected data.

[0036] Among them, ultrasonic detection technology is used to monitor the local stress changes inside the winding, including:

[0037] Acquire ultrasonic signal data inside the transformer winding through an ultrasonic sensor, the ultrasonic signal data including echo signals at different positions inside the winding;

[0038] Based on the preset ultrasonic signal eigenvalue template, the ultrasonic signal data is processed by eigenvalue to extract the eigenvalue related to the internal stress of the winding;

[0039] Compare the characteristic value of the ultrasonic signal with the preset reference characteristic value under normal working conditions to monitor the local stress changes inside the winding;

[0040] Based on the eigenvalues ​​of local stress changes and the eigenvalue template related to cuprous sulfide deposition, the physical changes in the winding are identified to determine whether there are abnormal physical changes caused by cuprous sulfide deposition.

[0041] Among them, spectral analysis technology is used to detect abnormal features in the spectrum, including:

[0042] Acquire the spectral radiation signal of the target to be measured, and acquire the abnormal characteristics in the spectral radiation signal;

[0043] Among them, the abnormal characteristics include the change characteristics of infrared band and ultraviolet band;

[0044] Based on the changing characteristics of infrared bands, identify whether the target to be tested has signs of high temperature;

[0045] Based on the changing characteristics of the ultraviolet band, identify whether the target to be tested has signs of partial discharge;

[0046] Determining an abnormal state of the target to be tested based on the high temperature sign and the partial discharge sign;

[0047] Determine the risk level of the target to be tested based on the abnormal state;

[0048] If the risk level of the target to be measured is higher than the preset threshold, the corresponding warning information is generated.

[0049] Among them, the monitoring data to be analyzed include:

[0050] Obtain the environmental parameter type from the monitored environmental parameters, and obtain the monitoring parameter type from the monitoring parameters;

[0051] Obtain the monitoring parameter type associated with each environmental parameter type as a data group to be analyzed;

[0052] After sorting each data group to be analyzed according to the environmental parameter type, the data corresponding to each environmental parameter type and monitoring parameter type is stored in the corresponding data group to be analyzed to obtain the monitoring data to be analyzed.

[0053] Among them, the deep learning analysis model is obtained, including:

[0054] When any parameter information in the data feature vector meets the preset conditions, the data feature vector is extracted as the initial data to be processed;

[0055] Perform cluster analysis on the initial data to be processed to obtain multiple clustering results;

[0056] According to the clustering result, a data group to be trained is determined, wherein the data group to be trained includes a multi-dimensional parameter feature vector;

[0057] Input the data group to be trained into a preset initial model, and perform model training on the data group to be trained based on the initial model;

[0058] During the model training process, deep learning technology is used to extract features and assign weights to multi-physical parameter monitoring data to obtain a trained deep learning analysis model;

[0059] Obtain the fault analysis results output by the deep learning analysis model and compare and verify them with the actual fault data;

[0060] If the analysis results meet the preset fault feature standards, the reliability of the deep learning analysis model is determined; otherwise, the training parameters of the deep learning analysis model are adjusted and retrained.

[0061] Among them, automatically processing and analyzing the collected data, including:

[0062] Construct data feature vectors, and perform matching analysis on the data feature vectors according to the preset fault feature library to identify discharge trends;

[0063] Evaluate the severity of discharge defects and generate fault diagnosis results;

[0064] Generate a comprehensive diagnostic report based on the fault diagnosis results, which includes discharge trends, severity assessment, and prevention and maintenance recommendations;

[0065] Comprehensive diagnostic reports are fed back to relevant personnel so that preventive and repair measures can be taken in a timely manner.

[0066] Compared with the prior art, the present invention has the following advantages:

[0067] The intelligent monitoring system for transformer windings based on cuprous sulfide deposition characteristics includes: a monitoring setting unit, which is used to create an online monitoring platform related to cuprous sulfide deposition characteristics in the target area; an acquisition unit, which is used to acquire multi-physical parameter monitoring data of transformer windings based on the online monitoring platform, and the monitoring data includes electromagnetic parameters, ultrasonic characteristics, light radiation characteristics, temperature and pressure information; an analysis unit, which is used to perform multi-dimensional analysis on multi-physical parameter monitoring data based on a deep learning analysis model to determine the potential discharge defects in transformer windings; and a diagnosis unit, which is used to automatically trigger an early warning when a potential discharge defect is detected, and collect all relevant monitoring data in the target area to form a comprehensive diagnosis report. This greatly improves the maintenance efficiency of the equipment and avoids significant economic losses caused by failure delays.

[0068] Other features and advantages of the present invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the present invention.

[0069] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0071] Figure 1 It is a structural diagram of a transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics in an embodiment of the present invention;

[0072] Figure 2 This is a structural diagram of a monitoring setting unit in an embodiment of the present invention;

[0073] Figure 3 This is a diagram of a unit structure obtained in an embodiment of the present invention. DETAILED DESCRIPTION

[0074] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0075] The embodiment of the present invention provides a transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics, including:

[0076] A monitoring setting unit for creating an online monitoring platform related to the characteristics of cuprous sulfide deposition in the target area;

[0077] An acquisition unit is used to acquire multi-physical parameter monitoring data of transformer windings based on an online monitoring platform, the monitoring data including electromagnetic parameters, ultrasonic characteristics, optical radiation characteristics, temperature and pressure information;

[0078] An analysis unit, which is used to perform multi-dimensional analysis on multi-physical parameter monitoring data based on a deep learning analysis model to determine potential discharge defects in transformer windings;

[0079] The diagnostic unit is used to automatically trigger an early warning when a potential discharge defect is detected, and to collect all relevant monitoring data in the target area to form a comprehensive diagnostic report.

[0080] The working principle of the above technical solution is as follows: the monitoring setting unit is used to create an online monitoring platform for the target area (such as transformer windings), which is specifically used to monitor the copper sulfide (Cu 2 S) parameters related to deposition characteristics. Cuprous sulfide is a conductive deposit formed in high-voltage electrical equipment, which causes internal discharge defects in transformers and leads to equipment failures. When the transformer is running, cuprous sulfide gradually accumulates on the surface of the winding. When the monitoring setting unit is started, the online platform will continuously collect and analyze the cuprous sulfide deposition in the target area and identify its growth pattern.

[0081] The acquisition unit is based on the online monitoring platform and is responsible for collecting multi-physical parameter monitoring data of the transformer windings. These data cover physical properties of different dimensions such as electromagnetic parameters, ultrasonic characteristics, light radiation characteristics, temperature and pressure, helping to comprehensively analyze the operating status of the equipment. While detecting cuprous sulfide deposition, the acquisition unit collects ultrasonic vibration data through a sensor network. When cuprous sulfide deposition causes a decrease in electrical insulation performance, a weak partial discharge signal is generated. These data can be used to detect signs of faults in advance.

[0082] The analysis unit uses a deep learning analysis model to perform multi-dimensional analysis on the acquired multi-physical parameter data. The deep learning model can automatically extract features from the data and identify potential abnormal patterns, such as discharge defects that may exist in the windings. The system integrates the collected temperature changes, electromagnetic field fluctuations, and light radiation characteristics to form a feature vector and input it into the deep learning model. The model analyzes that certain feature combinations are highly correlated with potential discharge defects, thereby concluding that there is partial discharge in the transformer windings.

[0083] When the analysis unit detects a possible discharge defect, the diagnostic unit automatically triggers an early warning. Subsequently, the unit summarizes all relevant monitoring data and generates a comprehensive diagnostic report with detailed analysis. This report helps maintenance personnel quickly locate the source of the fault and take appropriate measures. Once the diagnostic unit triggers an early warning, the system automatically generates a report that includes the results of multi-physical parameter analysis, identification of potential discharge locations, and possible solution suggestions. This greatly shortens the fault response time.

[0084] The beneficial effects of the above technical solution are: by creating a special online monitoring platform, the state of the transformer winding, especially the characteristics related to cuprous sulfide deposition, can be monitored in real time and continuously. This platform-based monitoring method improves the timeliness of data acquisition and the accuracy of monitoring. Since the system can obtain a variety of physical parameters (such as electromagnetic, ultrasonic, optical radiation, temperature, pressure, etc.), it provides a multi-dimensional panoramic view of the equipment status. This method avoids the misjudgment that may be caused by a single parameter, thereby improving the reliability of fault detection. The use of deep learning models to automatically analyze complex data reduces the need for human intervention, allowing the system to autonomously and quickly identify potential discharge defects. This not only improves the efficiency of analysis, but also improves the accuracy of fault identification. By automatically triggering early warnings and generating diagnostic reports, the system shortens the fault response time. This instant diagnosis mechanism greatly improves the maintenance efficiency of equipment and avoids significant economic losses caused by fault delays.

[0085] In another embodiment, the monitoring setting unit includes:

[0086] The platform construction module is used to build a real-time online monitoring platform in the target area, integrate various monitoring modules, and form a monitoring network covering various areas and key parts of the transformer winding;

[0087] Module integration module for integrating different monitoring modules related to cuprous sulfide deposition characteristics into an online monitoring platform;

[0088] The monitoring control module is used to coordinate different monitoring modules, perform joint monitoring operations on the target area, and obtain real-time monitoring data of cuprous sulfide deposition.

[0089] The working principle of the above technical solution is as follows: the platform construction module is responsible for building a real-time online monitoring platform in the target area (such as transformer windings). The platform integrates multiple monitoring modules to form a monitoring network covering the entire winding area, which can effectively monitor different areas and key parts of the transformer (such as high-voltage terminals and the core of the winding). For example: in a large transformer, the platform construction module will arrange multiple sensors at each layer and each key connection point of the winding. These sensors are integrated through the network to ensure that the monitoring platform can collect and transmit electromagnetic fields, temperature and vibration data inside the winding in real time. This multi-point monitoring can provide more accurate information on the health status of the equipment.

[0090] The main function of the module integration module is to integrate multiple monitoring modules related to cuprous sulfide deposition characteristics into the online monitoring platform. These monitoring modules are responsible for monitoring different physical parameters, such as electromagnetic fields, ultrasonic signals, light radiation or temperature changes. For example: in actual operation, cuprous sulfide deposition not only affects the electrical insulation performance of the transformer, but also causes local overheating of the winding. The module integration module will be responsible for integrating the sensor for monitoring temperature and the sensor for monitoring ultrasonic signals into the platform, so that the monitoring platform can capture both the temperature effect of deposition and the ultrasonic characteristics generated by partial discharge.

[0091] The monitoring control module coordinates the various monitoring modules to ensure that they can work together effectively and jointly monitor the target area. It can adjust the monitoring frequency, sensor sensitivity and other parameters of each module as needed, and obtain real-time monitoring data of cuprous sulfide deposition. For example, when the monitoring control module detects a rise in the local temperature of the winding, it can immediately increase the monitoring frequency of the electromagnetic field in that area to more accurately assess whether cuprous sulfide deposition has caused partial discharge. Through this joint monitoring, the system can achieve dynamic adjustment and improve the sensitivity and accuracy of monitoring.

[0092] The beneficial effects of the above technical solution are as follows: the platform construction module enables the monitoring system to cover all areas of the transformer winding, especially the key parts. This fully covered network can effectively capture any local abnormal phenomena, avoid monitoring blind spots, and ensure the safety of equipment operation. Through the module integration module, different monitoring modules can work together to monitor the various characteristics of cuprous sulfide deposition. In this way, not only can the changes in electrical parameters be monitored, but also the changes in parameters such as temperature, pressure and ultrasound can be monitored at the same time, so as to fully grasp the status of the equipment. The monitoring and control module can coordinate various modules, realize real-time data acquisition, and perform joint monitoring operations. When the monitoring results of a certain part are abnormal, the system can immediately adjust the working mode of other modules for more accurate monitoring. This not only improves the monitoring efficiency, but also enhances the early warning capability of the system. The monitoring and control module has flexibility and intelligent control capabilities, and can dynamically adjust the monitoring strategy according to the changes in real-time data. This mechanism can detect fault signs at an early stage and avoid the expansion of faults due to inadequate monitoring.

[0093] In another embodiment, the acquiring unit includes:

[0094] The electromagnetic parameter monitoring module is used to obtain the electromagnetic parameters of the transformer winding, including current fluctuations and voltage mutations, to identify discharge phenomena. The electromagnetic parameter monitoring module includes:

[0095] A signal extraction submodule, used to extract current and voltage signals from real-time operation data of transformer windings;

[0096] The feature analysis submodule is used to perform frequency domain and time domain analysis on the extracted current and voltage signals to identify abnormal fluctuations and mutation features.

[0097] The ultrasonic characteristic monitoring module is used to monitor the local stress changes inside the winding using ultrasonic detection technology and identify the physical changes caused by cuprous sulfide deposition, wherein the ultrasonic characteristic monitoring module includes:

[0098] An ultrasonic transmitting submodule, used for transmitting ultrasonic signals into the transformer winding;

[0099] An ultrasonic receiving submodule, used for receiving ultrasonic signals fed back from the winding;

[0100] The stress analysis submodule is used to process the received ultrasonic signal and analyze the stress changes inside the winding.

[0101] The optical radiation characteristic monitoring module is used to detect abnormal characteristics in the spectrum using spectral analysis technology, and the abnormal characteristics include changes in infrared and ultraviolet rays to identify signs of high temperature and partial discharge, wherein the optical radiation characteristic monitoring module includes:

[0102] An optical signal acquisition submodule, used to collect optical radiation signals around the transformer winding;

[0103] The spectrum analysis submodule is used to perform spectrum analysis on the collected optical signals and detect abnormal spectrum features.

[0104] The temperature and pressure information monitoring module is used to measure the transformer oil temperature and internal pressure, and analyze the winding temperature increase and pressure changes caused by cuprous sulfide deposition. The temperature and pressure information monitoring module includes:

[0105] Temperature sensor submodule, used to measure transformer oil temperature and winding temperature in real time;

[0106] Pressure sensing submodule, used to monitor the pressure changes inside the transformer;

[0107] The data correction submodule is used to correct and compensate the temperature and pressure data to ensure the measurement accuracy.

[0108] The data synchronization module is used to synchronize the electromagnetic parameters, ultrasonic characteristics, optical radiation characteristics, temperature and pressure information to ensure the coordination and consistency of multi-physical parameter monitoring data.

[0109] The working principle of the above technical solution is as follows: the electromagnetic parameter monitoring module is used to obtain the current fluctuation and voltage mutation of the transformer winding, and by analyzing these electromagnetic parameters, the possible discharge phenomenon of the transformer winding is identified. Signal extraction submodule: responsible for extracting current and voltage signals from the real-time operation data of the transformer winding. These signals can reflect the operating status of the winding. For example, when the transformer load suddenly increases, the current will fluctuate instantaneously. The signal extraction submodule captures these subtle fluctuations by analyzing the real-time changes of the current. Feature analysis submodule: frequency domain and time domain analysis of the extracted current and voltage signals. The frequency characteristics of the signal can be captured by frequency domain analysis, while the time domain analysis identifies the instantaneous changes of current or voltage. When partial discharge occurs inside the transformer, the voltage signal usually shows a short and violent mutation. The feature analysis submodule can identify these mutation characteristics through time domain analysis to determine whether there is a discharge phenomenon.

[0110] The ultrasonic characteristic monitoring module uses ultrasonic detection technology to monitor the local stress changes inside the transformer windings, thereby identifying the physical changes caused by cuprous sulfide deposition. Ultrasonic transmission submodule: responsible for sending ultrasonic signals into the windings. These signals can penetrate the transformer windings and can propagate inside the windings. If cuprous sulfide deposition intensifies in a certain area, the ultrasonic signal will be reflected by the local stress changes when passing through the area, causing the signal characteristics to change. Ultrasonic receiving submodule: Receives ultrasonic signals fed back from the windings. The received signals will be passed to the stress analysis submodule for processing. Stress analysis submodule: By analyzing the ultrasonic signals fed back, it is determined whether there are stress changes inside the windings, and then the physical effects of cuprous sulfide deposition are identified. If a winding is affected by stress at a certain location, the ultrasonic signal fed back will show a change in reflection intensity, indicating that there may be physical changes in the area, such as local damage or deposition.

[0111] The optical radiation characteristic monitoring module detects abnormal features in the spectrum through spectral analysis technology and identifies signs of high temperature and partial discharge. Optical signal acquisition submodule: used to collect optical radiation signals around the transformer windings. Abnormal optical radiation signals indicate local high temperature or discharge. Spectral analysis submodule: Spectral analysis is performed on the collected optical signals to identify abnormal spectral features, including changes in infrared and ultraviolet rays. If the temperature inside the transformer rises abnormally, the intensity of the infrared spectrum will increase. The spectral analysis submodule can identify the high temperature area by analyzing the changes in the infrared spectrum, helping to monitor abnormal winding temperature.

[0112] The temperature and pressure information monitoring module is responsible for measuring the transformer oil temperature and the temperature and pressure inside the winding, and identifying the temperature rise and pressure changes caused by cuprous sulfide deposition. Temperature sensing submodule: Real-time measurement of transformer oil temperature and winding temperature to ensure early detection of temperature anomalies. Pressure sensing submodule: Monitors pressure changes inside the transformer. A sharp increase in pressure indicates a fault inside the winding. For example: When a local temperature rise occurs somewhere in the transformer winding, the temperature sensor will immediately detect the temperature rise and report it to the system to help handle the fault in a timely manner. Data correction submodule: Correct and compensate the acquired temperature and pressure data to ensure measurement accuracy. This can eliminate errors caused by external factors such as ambient temperature fluctuations.

[0113] The data synchronization module is responsible for time synchronization of electromagnetic parameters, ultrasonic characteristics, optical radiation characteristics, temperature and pressure information to ensure that the data of all physical parameters are coordinated and consistent. When the ultrasonic characteristic monitoring module detects changes in internal stress of the winding, the data synchronization module will time match this information with the temperature and pressure information to provide consistent and accurate fault analysis.

[0114] The beneficial effects of the above technical solution are as follows: the electromagnetic parameter monitoring module can accurately identify early signs of electrical faults, such as partial discharge, by real-time monitoring of the current and voltage signals of the winding. Through dual analysis in the frequency domain and time domain, this monitoring method improves the accuracy of fault detection. The ultrasonic characteristic monitoring module can directly monitor the physical stress changes inside the winding, especially the local pressure changes caused by cuprous sulfide deposition. This ultrasonic detection technology can not only monitor in real time, but also cover areas that are difficult to detect in the winding, effectively preventing physical damage. Through the optical radiation characteristic monitoring module and the temperature monitoring module, local high temperature and discharge of the winding caused by cuprous sulfide deposition or other factors can be effectively identified. This combination of spectral analysis and temperature measurement provides a more comprehensive means of fault identification. The data synchronization module ensures the consistency of electromagnetic parameters, ultrasonic signals, spectral characteristics, temperature and pressure data, so that multiple monitoring modules can work in coordination and provide accurate joint fault diagnosis. Through the joint analysis of multiple physical parameters, the sensitivity and accuracy of fault detection can be effectively improved.

[0115] In another embodiment, determining a potential discharge defect existing in a transformer winding includes:

[0116] Identify and obtain current transformer monitoring data from the multi-physical parameter monitoring data, wherein the current transformer monitoring data includes monitoring parameters and monitoring environment parameters;

[0117] According to the monitoring parameters and monitoring environment parameters, the current transformer monitoring data is processed in layers to obtain the monitoring data to be analyzed;

[0118] Obtaining transformer fault standard data, and obtaining fault characteristic data from the transformer fault standard data;

[0119] After noise elimination processing is performed on the fault feature data, a data feature vector is constructed, and fault-related data is obtained from the transformer fault standard data;

[0120] According to the fault-related data, the data feature vectors are clustered to obtain a data group to be trained, and the initial model is trained with the data group to obtain a deep learning analysis model;

[0121] Input the monitoring data to be analyzed into the preset deep learning analysis model one by one to obtain the transformer status benchmark, including:

[0122] Inputting the first monitoring data to be analyzed into the deep learning analysis model, performing similarity matching on the first monitoring data to be analyzed in the deep learning analysis model, and outputting a first analysis result according to the matching result;

[0123] Starting from the second monitoring data to be analyzed, the previous analysis result is input into the deep learning analysis model to obtain the corresponding analysis result, and the first analysis result and all analysis results are integrated to obtain the transformer status benchmark;

[0124] The transformer status benchmark and current transformer monitoring data are input into the deep learning analysis model to obtain the monitoring result data of potential discharge defects in the transformer winding.

[0125] The working principle of the above technical solution is: extract the monitoring data related to the current transformer from the multi-physical parameter monitoring data. Transformer monitoring data includes two categories: Monitoring parameters: These are the operating parameters of the transformer, such as voltage, current, temperature, etc., which directly affect its state. Monitoring environmental parameters: These refer to factors related to the environment, such as humidity, external temperature, load conditions, etc.

[0126] Hierarchical processing of the extracted monitoring data means organizing and categorizing the data according to different categories or importance. Through this processing, the monitoring data to be analyzed can be filtered out more effectively, that is, the data most relevant to the transformer status or fault diagnosis.

[0127] Extract fault feature data related to specific failure modes from the transformer fault standard data. For example, if the standard data records historical failure cases of transformer windings, these data will contain common features of failures, such as partial discharge signals or abnormal winding temperature.

[0128] When processing these fault signature data, noise removal is first performed to ensure that the data used for analysis is accurate and clear. Noise refers to irrelevant information or errors in the data, such as errors in sensor readings or environmental interference. After noise removal, the data is converted into feature vectors, which can be processed more effectively in deep learning models.

[0129] Use the fault-related data to perform cluster analysis on the constructed feature vectors. Clustering refers to dividing data into different groups based on similarities, which can help the model better understand the differences between different fault modes. Through these clustered training data groups, an initial model is trained, and finally a deep learning analysis model capable of performing complex analysis is obtained.

[0130] The actual monitoring data to be analyzed is input into the deep learning model for step-by-step analysis. First, the first monitoring data is input, and the model will give the first analysis result based on the similarity matching result. Subsequently, all subsequent monitoring data will be combined with the previous analysis results during analysis to provide more continuous and dynamic fault monitoring.

[0131] After all the data is input and modeled, each analysis result is integrated to obtain the transformer status benchmark, which is used to indicate the overall health status of the current transformer or potential signs of failure.

[0132] Finally, the state baseline and the current transformer monitoring data are input into the deep learning model again to determine whether there are potential faults such as winding discharge defects. These monitoring results can help predict and prevent serious transformer failures.

[0133] The beneficial effects of the above technical solution are: by automatically processing and analyzing complex multi-physical parameter data through deep learning models, the limitations of traditional methods that rely on manual or simple algorithms are avoided, and potential problems can be discovered more quickly. The application of hierarchical processing and noise elimination technology ensures the high quality of data, making the analysis results more accurate. This high-precision monitoring can provide a more reliable health assessment for the transformer. Through the training and self-learning ability of the deep learning model, the system can improve the accuracy of diagnosis based on historical fault data and gradually optimize the model performance. Each analysis is based on the previous analysis results, ensuring dynamic monitoring of the transformer status. This continuity greatly improves the ability to track fault evolution. Early detection of faults can avoid major damage to the transformer during operation, thereby reducing downtime and maintenance costs.

[0134] In another embodiment, a comprehensive diagnostic report is formed, including:

[0135] Set dynamic thresholds. When monitoring parameters exceed the preset normal range, the monitoring result data of potential discharge defects will automatically trigger an early warning.

[0136] After the early warning is triggered, all relevant monitoring data are automatically collected in the target area to form a data set to be analyzed;

[0137] The data set to be analyzed is input into the deep learning analysis model to automatically process and analyze the collected data.

[0138] The working principle of the above technical solution is as follows: The dynamic threshold is a threshold setting method that is automatically adjusted according to real-time data. Compared with a fixed threshold, the dynamic threshold can be adjusted in real time according to changes in the environment or equipment status. For example, for transformer temperature monitoring, if the external ambient temperature rises, the dynamic threshold will also rise accordingly, thereby avoiding false alarms caused by changes in external factors. When the monitoring parameters (such as current, voltage, temperature, etc.) exceed this dynamic threshold, the system will determine that there may be potential discharge defects, thereby triggering an early warning.

[0139] When a parameter exceeds the set dynamic threshold, the system automatically triggers the early warning mechanism. After the early warning mechanism is activated, the system will immediately automatically collect all relevant monitoring data in the target area of ​​the transformer (such as a specific transformer or substation), including temperature, humidity, current fluctuations, vibrations and other information. These collected data are called the data set to be analyzed. Assuming that the temperature of the transformer winding suddenly rises beyond the dynamic threshold, the system will immediately record this situation and collect the voltage fluctuation data, ambient temperature, load changes and other information related to the temperature change into the data set to be analyzed.

[0140] The system inputs the collected data set to be analyzed into a pre-trained deep learning model. The model has been trained with a large amount of historical data and can identify different types of discharge defects. The deep learning model will first pre-process the input data, such as removing noise and standardizing, to ensure the consistency of the data during analysis. Based on the internal classification and prediction algorithms, the model analyzes the data to determine whether the current transformer has potential discharge defects, such as partial discharge, winding aging or insulation degradation.

[0141] After the analysis is completed, the system will automatically generate an analysis report, including the health status of the transformer and possible fault types, and give corresponding warning information. This information can be used by operation and maintenance personnel to quickly locate problems and conduct further inspections and maintenance.

[0142] The beneficial effects of the above technical solution are as follows: the application of dynamic thresholds ensures that the system can automatically adjust the warning standards according to real-time data changes, thereby improving the sensitivity and timeliness of fault detection. The process of automatically collecting relevant data after the warning is triggered avoids the trouble of manual screening of data, making the data collection process more efficient. Automated data processing and analysis reduces manpower input and improves the speed of fault handling. By analyzing the collected data through a deep learning model, the system can use the self-learning ability of the model to continuously optimize the accuracy of fault detection. For example, the model can gradually learn and identify the characteristics of discharge defects in different periods and environments, so as to give more accurate diagnostic results in future detections.

[0143] In another embodiment, the ultrasonic detection technology is used to monitor the local stress change inside the winding, including:

[0144] Acquire ultrasonic signal data inside the transformer winding through an ultrasonic sensor, the ultrasonic signal data including echo signals at different positions inside the winding;

[0145] Based on the preset ultrasonic signal eigenvalue template, the ultrasonic signal data is processed by eigenvalue to extract the eigenvalue related to the internal stress of the winding;

[0146] Compare the characteristic value of the ultrasonic signal with the preset reference characteristic value under normal working conditions to monitor the local stress changes inside the winding;

[0147] Based on the eigenvalues ​​of local stress changes and the eigenvalue template related to cuprous sulfide deposition, the physical changes in the winding are identified to determine whether there are abnormal physical changes caused by cuprous sulfide deposition.

[0148] The working principle of the above technical solution is as follows: the ultrasonic sensor is installed on the outside of the transformer winding, and the ultrasonic wave penetrates the winding material and obtains the echo signal inside the winding. The echo signal is the result of the ultrasonic wave being reflected when it encounters different materials or structural changes while propagating inside the winding. These signals can reflect the internal physical state of the winding. The ultrasonic wave is emitted from one end of the winding, and when it encounters material changes (such as stress concentration points or defects in the winding), a reflected echo is generated. The echo signals at different positions can show the different physical properties inside the winding.

[0149] After the ultrasonic signal is acquired, it needs to be processed based on the preset ultrasonic signal eigenvalue template. Eigenvalue is a process of converting the original signal into quantifiable indicators that are closely related to physical states (such as stress). During the processing, the system extracts key eigenvalues ​​related to the stress changes inside the winding, such as amplitude, frequency, signal attenuation rate, etc. The frequency change of a certain echo signal can reflect the change in the tightness of the winding material, and a significant increase in amplitude may indicate local stress concentration.

[0150] The extracted eigenvalues ​​will be compared with the baseline eigenvalues ​​of the windings under normal working conditions. The baseline eigenvalues ​​are pre-measured values ​​during transformer design and initial installation, reflecting the stress and structural characteristics of the windings under healthy conditions. If the eigenvalue at a certain location is significantly higher than the baseline value, it means that the stress state at that location may have changed, and there may be local stress concentration or material deformation in the windings.

[0151] Cuprous sulfide deposition is a common cause of transformer failure, which can cause electrical performance degradation or even failure within the winding. By comparing the characteristic value of local stress changes with the characteristic value template of cuprous sulfide deposition, it is possible to identify whether there are physical changes inside the winding caused by cuprous sulfide deposition. Cuprous sulfide deposition usually changes the surface characteristics of the winding and affects the propagation of ultrasonic signals, which manifests as abnormal signal attenuation or echo changes. If the system detects that the signal attenuation characteristics at a certain location match the template characteristic value of cuprous sulfide deposition, the system can identify that cuprous sulfide deposition may exist at that location and issue a relevant warning.

[0152] The beneficial effects of the above technical solution are as follows: through dynamic monitoring and eigenvalue processing of ultrasonic signals, the system can realize real-time monitoring of the internal stress state of the transformer winding. This technology can timely capture local stress concentration or structural changes inside the winding, thereby preventing small problems from evolving into major faults. Through eigenvalue processing and comparison with benchmark eigenvalues, this method can accurately identify changes in the physical state inside the winding. Ultrasonic signals can penetrate deep into the winding and detect subtle changes that are difficult to detect with traditional methods, thereby improving the accuracy of fault detection. Cuprous sulfide deposition is a major factor in transformer failure, and early identification of such deposition phenomena can effectively extend the service life of the transformer. By comparing with the cuprous sulfide deposition eigenvalue template, the system can issue an early warning before the deposition causes a major fault, preventing the fault from further deteriorating.

[0153] In another embodiment, a spectrum analysis technique is used to detect abnormal features in a spectrum, including:

[0154] Acquire the spectral radiation signal of the target to be measured, and acquire the abnormal characteristics in the spectral radiation signal;

[0155] Among them, the abnormal characteristics include the change characteristics of infrared band and ultraviolet band;

[0156] Based on the changing characteristics of infrared bands, identify whether the target to be tested has signs of high temperature;

[0157] Based on the changing characteristics of the ultraviolet band, identify whether the target to be tested has signs of partial discharge;

[0158] Determining an abnormal state of the target to be tested based on the high temperature sign and the partial discharge sign;

[0159] Determine the risk level of the target to be tested based on the abnormal state;

[0160] If the risk level of the target to be measured is higher than the preset threshold, the corresponding warning information is generated.

[0161] The working principle of the above technical solution is to capture the spectral radiation signal emitted by the target to be measured through a special spectral sensor or detection equipment. The spectral radiation signal is the light of different bands emitted by the target to be measured due to physical phenomena such as thermal radiation and electromagnetic radiation, such as visible light, infrared and ultraviolet light. In equipment such as power transformers, the band range of the radiation signal is stable during normal operation, but when the equipment is overheated or partially discharged, the radiation signals of different bands will produce abnormal changes.

[0162] The system analyzes the collected spectral radiation signals and extracts abnormal features. Abnormal features are changes in the spectral signals relative to the normal working state of the equipment, and these changes are usually concentrated in the infrared and ultraviolet bands. In the infrared band of a certain device, if the radiation intensity increases abnormally, it may indicate that the device is overheating; in the ultraviolet band, abnormal discharge may indicate that the device has electrical faults or insulation damage.

[0163] The system performs characteristic analysis on the infrared bands emitted by the target to be tested. Infrared radiation is usually directly related to the temperature of the target. If the radiation intensity of the infrared band increases beyond the normal value, the system can identify signs of high temperature in the equipment. When the temperature inside the power transformer rises, more infrared radiation will be generated, which is characterized by an increase in the peak value of the band. The system can judge that there is a risk of local overheating inside the equipment.

[0164] The UV band is mainly related to discharge phenomena in electrical equipment, especially when the insulation is damaged or the electrical contact is poor, which may cause tiny arcs or corona phenomena. By detecting the radiation changes in the UV band, the system can identify whether there are signs of partial discharge in the equipment. When partial discharge occurs in power equipment, the spectral signal of the UV band will show specific frequency changes and increased intensity, and the system will determine that the equipment has the potential for electrical discharge failure.

[0165] The system conducts a comprehensive analysis of the signs of high temperature in the infrared band and the signs of partial discharge in the ultraviolet band. The combination of the two can provide a more comprehensive description of the abnormal state of the target to be tested. For example, if the equipment is overheated and partially discharged at the same time, it means that the equipment is in a more dangerous state. If the equipment is detected to have both infrared overheating and ultraviolet discharge, the system can determine that the abnormal state of the equipment is very serious and requires immediate maintenance.

[0166] Based on the identified abnormal conditions, the system determines the risk level of the target to be tested. The risk level is determined based on preset threshold standards, which are determined according to the characteristics and operating conditions of different devices. If the abnormal condition exceeds the set threshold, the system determines the risk level as high. For example: when the overheating characteristics of infrared light exceed 50 degrees Celsius and the signs of ultraviolet discharge are obvious, the system can set the risk level of the device to "high risk".

[0167] If the risk level of the target exceeds the preset threshold, the system will automatically generate an early warning message and notify the relevant maintenance personnel for further inspection and processing. The early warning message contains the specific band characteristics of the abnormality, the risk level and the recommended maintenance operation. When a transformer is detected to have high temperature and partial discharge at the same time, the system will generate an emergency warning and recommend that the equipment be shut down immediately and inspected.

[0168] The beneficial effects of the above technical solution are as follows: by analyzing the radiation changes in the infrared and ultraviolet bands, the system can accurately capture the temperature anomalies and electrical fault phenomena of the target to be measured. This multi-band joint analysis method improves the accuracy of fault detection and can detect potential problems at an early stage. The system can realize real-time monitoring of the equipment and automatically generate early warning information when the abnormal state exceeds the preset threshold. This automated early warning mechanism reduces the pressure of manual monitoring, ensures timely response to potential risks of the equipment, and reduces the possibility of sudden equipment failure. The method can effectively identify high temperature and partial discharge phenomena in the equipment, helping maintenance personnel to deal with the problem before it worsens. This early warning mechanism can extend the service life of the equipment and avoid serious damage caused by overheating or discharge.

[0169] In another embodiment, obtaining the monitoring data to be analyzed includes:

[0170] Obtain the environmental parameter type from the monitored environmental parameters, and obtain the monitoring parameter type from the monitoring parameters;

[0171] Obtain the monitoring parameter type associated with each environmental parameter type as a data group to be analyzed;

[0172] After sorting each data group to be analyzed according to the environmental parameter type, the data corresponding to each environmental parameter type and monitoring parameter type is stored in the corresponding data group to be analyzed to obtain the monitoring data to be analyzed.

[0173] The working principle of the above technical solution is as follows: first, environmental parameters are collected from the monitoring system. These environmental parameters refer to specific indicators or elements related to the environment to be measured. For example, temperature, humidity, atmospheric pressure, wind speed, etc. are all environmental parameters. Each environmental parameter has a corresponding environmental parameter type, which is used to describe the nature of the parameter.

[0174] In the process of environmental monitoring, in addition to environmental parameters, monitoring parameters are also obtained, that is, values ​​related to specific monitoring items of the target or environment to be measured, such as the operating temperature, vibration frequency, current, voltage, etc. of the equipment. These monitoring parameters have their own monitoring parameter types to describe their characteristics.

[0175] The system associates each environmental parameter type with the relevant monitoring parameter type through preset rules or based on some logic. This association process is matched according to the relationship between environmental factors and monitored objects. Each environmental parameter type corresponds to one or more monitoring parameter types, forming a data group to be analyzed.

[0176] After obtaining the associated environmental parameter type and monitoring parameter type, the system sorts the data in each data group to be analyzed. The sorting can be based on time order, numerical value, or other analysis-related criteria to ensure that the data has logical consistency in subsequent analysis. If you want to analyze the relationship between device temperature and ambient temperature, you can sort the two types of data in time order to ensure that they correspond to the same time point.

[0177] After sorting, the system stores the associated data corresponding to each environmental parameter type and monitoring parameter type into the corresponding data group to be analyzed. Each data group contains matching data of environmental parameters and monitoring parameters, ready for subsequent analysis. The sorted data of temperature environmental parameters and device temperature monitoring parameters are stored in the same data group, and the data of humidity parameters and resistance parameters are stored in another data group.

[0178] Finally, the system forms the monitoring data to be analyzed based on the sorted and associated data. These data are organized according to the correspondence between specific environmental parameters and monitoring parameters, which is convenient for subsequent data analysis and trend prediction. After the system completes the data association, it generates a data set including "temperature-equipment operating temperature" and "humidity-insulation resistance" to be analyzed.

[0179] The beneficial effects of the above technical solution are as follows: by associating environmental parameter types with monitoring parameter types, the system can ensure that data from different sources are accurately matched, and this association can help analyze the complex relationship between environmental conditions and monitored objects. The sorting and storage process of data groups makes subsequent analysis more efficient. Sorting ensures the consistency of data in the time dimension or numerical dimension, avoids chaotic data input, and reduces the time for data cleaning. Analysts can quickly obtain ordered and clearly associated data sets, thereby improving the speed and efficiency of analysis. Allowing multiple environmental parameter types to be associated with multiple monitoring parameter types provides a basis for comprehensive analysis. Through the combined analysis of multiple environmental and monitoring data, more potential trends and correlations can be discovered.

[0180] In another embodiment, obtaining a deep learning analysis model includes:

[0181] When any parameter information in the data feature vector meets the preset conditions, the data feature vector is extracted as the initial data to be processed;

[0182] Perform cluster analysis on the initial data to be processed to obtain multiple clustering results;

[0183] According to the clustering result, a data group to be trained is determined, wherein the data group to be trained includes a multi-dimensional parameter feature vector;

[0184] Input the data group to be trained into a preset initial model, and perform model training on the data group to be trained based on the initial model;

[0185] During the model training process, deep learning technology is used to extract features and assign weights to multi-physical parameter monitoring data to obtain a trained deep learning analysis model;

[0186] Obtain the fault analysis results output by the deep learning analysis model and compare and verify them with the actual fault data;

[0187] If the analysis results meet the preset fault feature standards, the reliability of the deep learning analysis model is determined; otherwise, the training parameters of the deep learning analysis model are adjusted and retrained.

[0188] The working principle of the above technical solution is: when any parameter in a data set meets the pre-set conditions, the data feature vector will be extracted as the initial data to be processed. For example, in the status monitoring of equipment operation, if physical parameters such as temperature or pressure exceed the normal range, these feature parameters will be identified and extracted for subsequent analysis.

[0189] The initial data to be processed will be subjected to cluster analysis, which is an unsupervised learning method in machine learning that aims to group data based on similarity. For example, assuming we have a set of industrial sensor data, cluster analysis will group data with similar parameters such as temperature, pressure, and vibration into one category to help discover failure modes with potential similar characteristics. Commonly used clustering algorithms include K-means clustering, DBSCAN, etc.

[0190] Based on the clustering results, the system will select a set of data containing multi-dimensional parameter feature vectors as the training data. These parameter dimensions can include physical parameters such as temperature, pressure, and vibration frequency. Through this multi-dimensional data input, more comprehensive information can be provided to the model to help the model better learn the failure mode.

[0191] The data set to be trained is input into a preset initial model, which can be a deep neural network model (DNN) or other suitable machine learning model. At this stage, the model is trained based on the input data to learn the relationship between various features so that prediction or classification can be performed.

[0192] During the training process, the system uses deep learning technology to extract features and assign weights to multiple physical parameters. For example, in fault analysis, the system will identify which features (such as temperature and vibration) contribute more to the fault and assign higher weights to these features. This process helps the model improve the accuracy of fault detection.

[0193] When the model training is completed, the deep learning analysis model will output a fault analysis result. This result will be compared and verified with the actual fault data. For example, if the vibration frequency of a certain device exceeds the normal range, the system will compare the predicted fault with the actual fault. Based on the comparison results, the accuracy and reliability of the model are determined.

[0194] If the analysis results do not meet the preset fault feature standards, the system will adjust the model's training parameters (such as learning rate, number of network layers, etc.) and retrain until the fault analysis results output by the model meet expectations.

[0195] The beneficial effects of the above technical solution are: through cluster analysis, the system can identify and group similar fault modes, reduce the redundancy of feature dimensions, and improve the classification accuracy of the model in different fault scenarios. At the same time, the application of deep learning technology can automatically extract important features from the data to ensure that the model can more accurately predict the operating status and potential faults of the equipment. By using feature extraction and cluster analysis of multiple physical parameters, the system can effectively screen out important feature vectors, reduce the training time of the model, and improve the efficiency of the entire fault detection process. This method can quickly process massive amounts of data and is suitable for scenarios that require real-time analysis, such as industrial monitoring and equipment maintenance.

[0196] In another embodiment, the collected data is automatically processed and analyzed, including:

[0197] Construct data feature vectors, and perform matching analysis on the data feature vectors according to the preset fault feature library to identify discharge trends;

[0198] Evaluate the severity of discharge defects and generate fault diagnosis results;

[0199] Generate a comprehensive diagnostic report based on the fault diagnosis results, which includes discharge trends, severity assessment, and prevention and maintenance recommendations;

[0200] Comprehensive diagnostic reports are fed back to relevant personnel so that preventive and repair measures can be taken in a timely manner.

[0201] The working principle of the above technical solution is as follows: In the discharge trend analysis, it is first necessary to extract multiple physical parameters from the monitoring data and construct a data feature vector. These parameters include current, voltage, discharge amount, temperature, etc. Suppose that when we monitor a high-voltage power device, we record its operating voltage, operating current, and temperature data of its internal sensors. By collating the measured values ​​of these different time periods, a set of multi-dimensional data feature vectors can be generated for subsequent analysis.

[0202] The system will match and analyze the extracted feature vectors according to the preset fault feature library. The fault feature library contains various known discharge modes and their characteristic parameters (for example, discharge frequency, discharge time interval, discharge amplitude change, etc.). Assume that there is a typical "intermittent partial discharge" mode in the feature library, which is characterized by a periodic increase in discharge current within a certain period of time and a sudden increase in discharge amplitude. The system will match the current feature vector with the mode. If the trend of the current data is similar to the mode, the system will identify the discharge trend and issue an alarm.

[0203] When the discharge trend is identified, the system will further evaluate the severity of the defect. The severity is usually based on parameters such as discharge frequency, discharge amplitude, and discharge duration. For example, if the discharge frequency gradually increases and the discharge amplitude exceeds a certain safety threshold (for example, the discharge current of a device gradually increases from 0.5mA to 1.5mA), the system will determine that the discharge defect is a serious state and mark it as a "high-risk fault" in the diagnostic report, indicating that it needs to be handled immediately.

[0204] Based on the evaluation results, the system will generate corresponding fault diagnosis results. The diagnosis results usually include: discharge trend (whether the discharge has an upward or downward trend), the severity of the discharge defect (minor, moderate or severe), and possible fault cause analysis (such as insulation aging, internal partial discharge, etc.). The system generates a detailed comprehensive diagnosis report based on the fault diagnosis results. The report will include the following aspects:

[0205] Discharge trend analysis: Describe in detail the changes in the discharge trend (for example, the growth trend of the discharge frequency, the fluctuation of the discharge amplitude, etc.).

[0206] Severity assessment: Based on the analysis results, the severity of the discharge defect is graded and its potential risks are explained.

[0207] Prevention and maintenance suggestions: Based on the diagnosis results, the system will provide some preventive measures and maintenance suggestions. For example, if the discharge amplitude is found to be increasing, it can be recommended that relevant personnel conduct insulation testing or replace certain key components.

[0208] Finally, a comprehensive diagnostic report is sent to the relevant maintenance personnel. Based on the fault trend and severity assessment provided in the report, the maintenance personnel can take preventive measures in a timely manner. For example, if the report indicates that the discharge trend may reach a dangerous level in the next week, the equipment can be scheduled for downtime inspection or replacement of wearing parts in advance, thus preventing major accidents.

[0209] The beneficial effects of the above technical solution are: by establishing a detailed fault feature library and using matching analysis technology, the system can accurately identify discharge trends, thereby effectively avoiding misjudgments caused by complex discharge patterns. This precise analysis capability is particularly suitable for complex industrial equipment environments, such as status monitoring of high-voltage power equipment or large transformers. The fault diagnosis report not only includes the identification and evaluation of the current fault, but also provides detailed prevention and maintenance recommendations. This comprehensive analysis result can help maintenance personnel better understand the severity of the fault and develop more effective maintenance strategies. For example, the system can reduce operational risks by predicting discharge trends in advance to avoid sudden failures of equipment during high-load operation.

[0210] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. Transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics, characterized in that: include: A monitoring setting unit for creating an online monitoring platform related to the characteristics of cuprous sulfide deposition in the target area; An acquisition unit is used to acquire multi-physical parameter monitoring data of transformer windings based on an online monitoring platform, the monitoring data including electromagnetic parameters, ultrasonic characteristics, optical radiation characteristics, temperature and pressure information; An analysis unit, which is used to perform multi-dimensional analysis on multi-physical parameter monitoring data based on a deep learning analysis model to determine potential discharge defects in transformer windings; The diagnostic unit is used to automatically trigger an early warning when a potential discharge defect is detected, and collect all relevant monitoring data in the target area to form a comprehensive diagnostic report; Identify potential discharge defects in transformer windings, including: Identify and obtain current transformer monitoring data from the multi-physical parameter monitoring data, wherein the current transformer monitoring data includes monitoring parameters and monitoring environment parameters; According to the monitoring parameters and monitoring environment parameters, the current transformer monitoring data is processed in layers to obtain the monitoring data to be analyzed; Obtaining transformer fault standard data, and obtaining fault characteristic data from the transformer fault standard data; After noise elimination processing is performed on the fault feature data, a data feature vector is constructed, and fault-related data is obtained from the transformer fault standard data; According to the fault-related data, the data feature vectors are clustered to obtain a data group to be trained, and the initial model is trained with the data group to obtain a deep learning analysis model; Input the monitoring data to be analyzed into the preset deep learning analysis model one by one to obtain the transformer status benchmark, including: Inputting the first monitoring data to be analyzed into the deep learning analysis model, performing similarity matching on the first monitoring data to be analyzed in the deep learning analysis model, and outputting a first analysis result according to the matching result; Starting from the second monitoring data to be analyzed, the previous analysis result is input into the deep learning analysis model to obtain the corresponding analysis result, and the first analysis result and all analysis results are integrated to obtain the transformer status benchmark; Input the transformer status benchmark and current transformer monitoring data into the deep learning analysis model to obtain the monitoring result data of potential discharge defects in the transformer winding; Obtain monitoring data to be analyzed, including: Obtain the environmental parameter type from the monitored environmental parameters, and obtain the monitoring parameter type from the monitoring parameters; Obtain the monitoring parameter type associated with each environmental parameter type as a data group to be analyzed; After sorting each data group to be analyzed according to the environmental parameter type, the data corresponding to each environmental parameter type and monitoring parameter type is stored in the corresponding data group to be analyzed to obtain the monitoring data to be analyzed; Get the deep learning analysis model, including: When any parameter information in the data feature vector meets the preset conditions, the data feature vector is extracted as the initial data to be processed; Perform cluster analysis on the initial data to be processed to obtain multiple clustering results; According to the clustering result, a data group to be trained is determined, wherein the data group to be trained includes a multi-dimensional parameter feature vector; Input the data group to be trained into a preset initial model, and perform model training on the data group to be trained based on the initial model; During the model training process, deep learning technology is used to extract features and assign weights to multi-physical parameter monitoring data to obtain a trained deep learning analysis model; Obtain the fault analysis results output by the deep learning analysis model and compare and verify them with the actual fault data; If the analysis results meet the preset fault feature standards, the reliability of the deep learning analysis model is determined; otherwise, the training parameters of the deep learning analysis model are adjusted and retrained.

2. The intelligent transformer winding monitoring system based on cuprous sulfide deposition characteristics according to claim 1 is characterized in that: The monitoring setting unit includes: The platform construction module is used to build a real-time online monitoring platform in the target area, integrate various monitoring modules, and form a monitoring network covering various areas and key parts of the transformer winding; Module integration module for integrating different monitoring modules related to cuprous sulfide deposition characteristics into an online monitoring platform; The monitoring control module is used to coordinate different monitoring modules, perform joint monitoring operations on the target area, and obtain real-time monitoring data of cuprous sulfide deposition.

3. The transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics according to claim 1 is characterized in that: The acquisition unit includes: Electromagnetic parameter monitoring module, used to obtain electromagnetic parameters of transformer windings, including current fluctuations and voltage mutations, to identify discharge phenomena; Ultrasonic characteristic monitoring module, which is used to monitor the local stress changes inside the winding using ultrasonic detection technology and identify physical changes caused by cuprous sulfide deposition; An optical radiation characteristics monitoring module for detecting abnormal features in the spectrum using spectral analysis techniques, including changes in infrared and ultraviolet light, to identify signs of high temperature and partial discharge; Temperature and pressure information monitoring module, used to measure transformer oil temperature and internal pressure, and analyze winding temperature rise and pressure changes caused by cuprous sulfide deposition; The data synchronization module is used to synchronize the electromagnetic parameters, ultrasonic characteristics, optical radiation characteristics, temperature and pressure information to ensure the coordination and consistency of multi-physical parameter monitoring data.

4. The transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics according to claim 1 is characterized in that: Generate a comprehensive diagnostic report, including: Set dynamic thresholds. When monitoring parameters exceed the preset normal range, the monitoring result data of potential discharge defects will automatically trigger an early warning. After the early warning is triggered, all relevant monitoring data are automatically collected in the target area to form a data set to be analyzed; The data set to be analyzed is input into the deep learning analysis model to automatically process and analyze the collected data.

5. The transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics according to claim 3 is characterized in that: Ultrasonic testing technology is used to monitor local stress changes inside the winding, including: Acquire ultrasonic signal data inside the transformer winding through an ultrasonic sensor, the ultrasonic signal data including echo signals at different positions inside the winding; Based on the preset ultrasonic signal eigenvalue template, the ultrasonic signal data is processed by eigenvalue to extract the eigenvalue related to the internal stress of the winding; Compare the characteristic value of the ultrasonic signal with the preset reference characteristic value under normal working conditions to monitor the local stress changes inside the winding; Based on the eigenvalues ​​of local stress changes and the eigenvalue template related to cuprous sulfide deposition, the physical changes in the winding are identified to determine whether there are abnormal physical changes caused by cuprous sulfide deposition.

6. The transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics according to claim 3 is characterized in that: Use spectral analysis techniques to detect unusual features in the spectrum, including: Acquire the spectral radiation signal of the target to be measured, and acquire the abnormal characteristics in the spectral radiation signal; Among them, the abnormal characteristics include the change characteristics of infrared band and ultraviolet band; Based on the changing characteristics of infrared bands, identify whether the target to be tested has signs of high temperature; Based on the changing characteristics of the ultraviolet band, identify whether the target to be tested has signs of partial discharge; Determining an abnormal state of the target to be tested based on the high temperature sign and the partial discharge sign; Determine the risk level of the target to be tested based on the abnormal state; If the risk level of the target to be measured is higher than the preset threshold, the corresponding warning information is generated.

7. The transformer winding intelligent monitoring system based on cuprous sulfide deposition characteristics according to claim 4 is characterized in that: Automatically process and analyze collected data, including: Construct data feature vectors, and perform matching analysis on the data feature vectors according to the preset fault feature library to identify discharge trends; Evaluate the severity of discharge defects and generate fault diagnosis results; Generate a comprehensive diagnostic report based on the fault diagnosis results, which includes discharge trends, severity assessment, and prevention and maintenance recommendations; Comprehensive diagnostic reports are fed back to relevant personnel so that preventive and repair measures can be taken in a timely manner.

Images