Method and system for failure early warning of power transmission and distribution equipment

Abstract

The application provides a fault early warning method and system for power transmission and distribution equipment, and relates to the technical field of power transmission and distribution equipment. The method comprises the following steps: obtaining real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameter by monitoring the transformer in real time through a sensing module; periodically sampling the current transformer and voltage transformer of the power distribution network in the power system to obtain current-voltage three-phase data sequence; performing state analysis to trigger a fault early warning instruction; performing feature backtracking on the sensing module to generate a transformer feature backtracking result; performing trend feature identification on the current-voltage three-phase data sequence to generate current-voltage three-phase data trend features; and performing fault early warning analysis to obtain a fault early warning result. The application can solve the technical problem of low fault early warning accuracy of the transformer and power distribution equipment in the prior art, and achieve the technical effect of improving early warning accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power transmission and distribution equipment, and particularly relates to a fault early warning method and system for power transmission and distribution equipment. BACKGROUND

[0002] With the continuous expansion of the scale of the power system and the increasing complexity of the operation environment of power transmission and distribution equipment, the traditional equipment fault identification method relying on artificial experience or static rules has been difficult to meet the requirements of high-reliability power grid operation. After long-term operation of power transmission and distribution equipment, it will be affected by multiple factors such as electrical load, temperature environment, humidity, equipment aging, etc. to varying degrees, and gradually show a fault evolution trend with equipment individual characteristics. This trend feature is often not instantaneous mutation, but in the form of slow deterioration or periodic fluctuation, which is difficult to be captured and judged in time by traditional methods. At present, the traditional early warning of power transmission and distribution equipment mainly relies on setting thresholds to monitor single parameters, ignoring the linkage trend and historical evolution trajectory among different parameters. Moreover, in addition to real-time monitoring, the deep value of historical data is often not fully utilized, resulting in poor adaptability and high false alarm rate of the early warning model.

[0003] In summary, in the prior art, since the early warning of power transmission and distribution equipment mainly relies on setting thresholds to monitor single parameters, the linkage trend and historical evolution trajectory among different parameters are ignored, which further affects the fault early warning accuracy and timeliness of transformer and power distribution equipment, thereby weakening the rapid perception and risk control ability of the entire power grid under abnormal state. SUMMARY

[0004] The purpose of the present application is to provide a fault early warning method and system for power transmission and distribution equipment, which solves the technical problem in the prior art that the early warning of power transmission and distribution equipment mainly relies on setting thresholds to monitor single parameters, resulting in ignoring the linkage trend and historical evolution trajectory among different parameters, which further affects the fault early warning accuracy and timeliness of transformer and power distribution equipment, thereby weakening the rapid perception and risk control ability of the entire power grid under abnormal state.

[0005] In view of the above problems, the present application provides a fault early warning method and system for power transmission and distribution equipment.

[0006] In a first aspect, the application provides a fault early warning method for power transmission and distribution equipment, which is implemented by a fault early warning system for power transmission and distribution equipment, and includes the following steps: a perception module for health state monitoring of a transformer of a power system is constructed by using an oil chromatographic analyzer, a winding temperature sensor and a microcomputer monitoring device, the transformer is monitored in real time by the perception module to obtain real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameter; a feeder terminal unit is used to periodically sample current transformers and voltage transformers of a power distribution network in the power system according to a preset sampling period to obtain current-voltage three-phase data sequence; state analysis is performed based on the real-time gas concentration-generation rate, the real-time temperature rise monitoring value and the real-time integrated monitoring parameter, and the current-voltage three-phase data sequence to trigger a fault early warning instruction; feature backtracking of the perception module is performed based on the fault early warning instruction to generate a transformer feature backtracking result; trend feature identification is performed on the current-voltage three-phase data sequence to generate current-voltage three-phase data trend features; and fault early warning analysis is performed according to the transformer feature backtracking result and the current-voltage three-phase data trend features to obtain a fault early warning result.

[0007] Preferably, the fault early warning method for power transmission and distribution equipment further includes that the perception module has a data buffer area, and the data buffer area is used for data storage of a feature backtracking window.

[0008] Preferably, the fault early warning method for power transmission and distribution equipment further includes that backtracking data stored in the data buffer area of the perception module is called based on the fault early warning instruction to obtain a gas concentration-generation rate backtracking data sequence, a temperature rise monitoring value backtracking data sequence and an integrated monitoring parameter backtracking data sequence; the real-time gas concentration-generation rate, the real-time temperature rise monitoring value and the real-time integrated monitoring parameter are taken as foreground data, and the gas concentration-generation rate backtracking data sequence, the temperature rise monitoring value backtracking data sequence and the integrated monitoring parameter backtracking data sequence are taken as background data; mapping feature backtracking analysis is performed on the foreground data and the background data to generate a transformer feature backtracking result.

[0009] Preferably, the fault early warning method for power transmission and distribution equipment further comprises: extracting the real-time gas concentration-generation rate in the foreground data, mapping and deviation analyzing the gas concentration-generation rate backtracking data sequence in the background data, and obtaining a gas concentration-generation rate deviation value sequence; center value screening the gas concentration-generation rate deviation value sequence to determine a target backtracking center value; using the target backtracking center value to perform feature analysis data screening on the gas concentration-generation rate deviation value sequence, and performing trend feature analysis on the screening result to generate a gas concentration-generation rate backtracking feature; respectively performing mapping feature backtracking analysis on the foreground data and the background data to generate a temperature rise monitoring value backtracking feature and an integrated monitoring parameter backtracking feature, and combining the gas concentration-generation rate backtracking feature to obtain the transformer feature backtracking result.

[0010] Preferably, the fault early warning method for power transmission and distribution equipment further comprises: extracting the maximum deviation value in the gas concentration-generation rate deviation value sequence as an initial backtracking data center value; and performing deep embedded center value iteration on the gas concentration-generation rate deviation value sequence based on the initial backtracking data center value to determine a target backtracking data center value.

[0011] Preferably, the fault early warning method for power transmission and distribution equipment further comprises: pre-constructing a trend feature identification component, wherein the trend feature identification component comprises a multi-scale trend feature identification branch set and a multi-scale trend feature fusion branch; performing multi-scale analysis on the current-voltage three-phase data sequence using the multi-scale trend feature identification branch set to determine a multi-scale trend feature set; and performing interactive enhancement on the multi-scale trend feature set using the multi-scale trend feature fusion branch to obtain the current-voltage three-phase data trend feature.

[0012] Preferably, the fault early warning method for power transmission and distribution equipment further comprises: calling a feature similarity identification component in the multi-scale trend feature fusion branch; randomly extracting two multi-scale trend features from the multi-scale trend feature set without replacement, identifying the similarity of the two multi-scale trend features using the feature similarity identification component to obtain a feature similarity set; calling an interaction matrix construction component in the multi-scale trend feature fusion branch; normalizing the feature similarity set using the interaction matrix construction component, and filling the processed result into an initially empty matrix to obtain an interaction matrix; and respectively performing feature enhancement on the two multi-scale trend features using the interaction matrix to generate two enhanced multi-scale trend features; and iteratively obtaining a multi-scale enhanced trend feature set, and performing mean value processing on the multi-scale enhanced trend feature set to obtain the current-voltage three-phase data trend feature.

[0013] Preferably, the fault early warning method for power transmission and distribution equipment further comprises: determining a warning feedback verification window based on the fault early warning result; and troubleshooting and verifying the transformer and the power distribution network in the warning feedback verification window to obtain a verification result.

[0014] Preferably, the fault early warning method for power transmission and distribution equipment further comprises: adjusting the preset sampling period of the feeder terminal unit and the feature backtracking window of the perception module based on the verification result.

[0015] In a second aspect, the present application further provides a fault early warning system for power transmission and distribution equipment, configured to execute the fault early warning method for power transmission and distribution equipment as described in the first aspect, comprising: a perception module construction module configured to construct a perception module for health state monitoring of a transformer of a power system by using an oil chromatographic analyzer, a winding temperature sensor and a microcomputer monitoring device, to obtain real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameter by real-time monitoring of the transformer through the perception module; a periodic sampling module configured to periodically sample current transformers and voltage transformers of a power distribution network in the power system according to a preset sampling period of a feeder terminal unit to obtain current-voltage three-phase data sequence; a state analysis module configured to perform state analysis based on the real-time gas concentration-generation rate, the real-time temperature rise monitoring value, the real-time integrated monitoring parameter and the current-voltage three-phase data sequence to trigger a fault early warning instruction; a feature backtracking module configured to perform feature backtracking on the perception module based on the fault early warning instruction to generate a transformer feature backtracking result; a trend feature identification module configured to perform trend feature identification on the current-voltage three-phase data sequence to generate current-voltage three-phase data trend features; and a fault early warning analysis module configured to perform fault early warning analysis according to the transformer feature backtracking result and the current-voltage three-phase data trend features to obtain a fault early warning result.

[0016] The technical solutions provided in the present application have at least the following technical effects or advantages: by achieving the technical target of dynamic fault early warning analysis based on multi-source monitoring data fusion and trend feature evolution analysis, the technical effects of improving early warning accuracy, enhancing system adaptability, reducing false positive rate and false negative rate, achieving early fault identification and assisting decision support are achieved.

[0017] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented in accordance with the content of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become apparent from the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the fault early warning method for power transmission and distribution equipment used in this application.

[0020] Figure 2 This is a schematic diagram of the fault early warning system for power transmission and distribution equipment used in this application.

[0021] Figure labeling: Perception module construction module 11, periodic sampling module 12, state analysis module 13, feature backtracking module 14, trend feature recognition module 15, fault early warning analysis module 16. Detailed Implementation

[0022] This application provides a fault early warning method and system for power transmission and distribution equipment, solving the technical problem that existing technologies often rely on monitoring single parameters with set thresholds for early warning of power transmission and distribution equipment. This neglects the linkage trends and historical evolution trajectories between different parameters, further affecting the accuracy of fault early warning and the timeliness of early warning response for transformers and distribution equipment, thereby weakening the entire power grid's ability to quickly perceive and control risks under abnormal conditions. The application achieves the technical goal of dynamic fault early warning analysis based on multi-source monitoring data fusion and trend feature evolution analysis, thereby improving early warning accuracy, enhancing system adaptability, reducing false alarm and false negative rates, and enabling early fault identification and decision support.

[0023] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0024] Example 1, please refer to the appendix. Figure 1 This application provides a fault early warning method for power transmission and distribution equipment, which is applied to a fault early warning system for power transmission and distribution equipment, and specifically includes the following steps:

[0025] S1: Construct a perception module for health state monitoring of the transformer of the power system by using an oil chromatographic analyzer, a winding temperature sensor, and a microcomputer monitoring device. Real-time gas concentration-generation rate, real-time temperature rise monitoring value, and real-time integrated monitoring parameters are obtained by real-time monitoring of the transformer through the perception module.

[0026] Specifically, the health state of the transformer in the power system is monitored by constructing a perception module using an oil chromatographic analyzer, a winding temperature sensor, and a microcomputer monitoring device. The oil chromatographic analyzer is a device for detecting trace gases in transformer insulating oil, including hydrogen, carbon monoxide, acetylene, etc. When there is a partial discharge, overheating, or arc fault inside the transformer, different components and proportions of gases will be released in the oil. Analyzing the concentration and generation rate of the gases helps to determine the development trend of potential faults inside the transformer. The winding temperature sensor is used to detect the temperature of the transformer winding in real time. High temperature of the winding may cause insulation aging or even thermal breakdown. Therefore, continuous monitoring of the temperature rise is an important means to ensure the safe operation of the transformer. The microcomputer monitoring device is an integrated intelligent terminal that can collect and process multiple monitoring signals. It has local processing capability and supports remote uploading to the dispatching system. The microcomputer monitoring device can collect real-time electrical parameters such as three-phase voltage, current, active power, reactive power, frequency, and load current. It can also monitor the operating state such as the on-load voltage regulation position, the start-stop state of the fan and water pump, the oil level, the oil pressure, and the normal operation of the cooler. It can also detect the insulation state, such as partial discharge or leakage current. When a fault occurs, the microcomputer monitoring device will collect alarm data such as temperature overrun, gas exceeding the standard, arc generation, etc., and form an alarm record, which is uploaded to the SCADA (Supervisory Control and Data Acquisition System) or DCS (Distributed Control System) for technicians to carry out early warning and trend analysis.

[0027] Real-time gas concentration-generation rate, real-time temperature rise monitoring value, and real-time integrated monitoring parameters can be obtained through the perception module. The gas concentration-generation rate is obtained by the oil chromatographic analyzer, which represents the growth of a specific gas in the insulating oil per unit time, and can reflect whether a certain type of fault is developing rapidly. For example, if the hydrogen concentration increases from 100 μL / L to 300 μL / L in 2 hours, it may indicate the presence of arc discharge. The real-time temperature rise monitoring value is obtained by the winding temperature sensor, which refers to the temperature rise of the winding or oil relative to the baseline environment. For example, if the winding temperature rises by 30 degrees Celsius, it may reach the warning threshold. The real-time integrated monitoring parameters are obtained by the microcomputer monitoring device, which are used to comprehensively judge the equipment state. For example, three-phase voltage, current, active / reactive power, frequency, load current, on-load voltage regulation state, fan / water pump start-stop signal, oil level, oil pressure, cooler operating state, etc. Table 1 shows the parameter record of the latest real-time integrated monitoring parameters.

[0028] Table 1: Parameter record of real-time integrated monitoring parameters collected last time

[0029]

[0030] S2: According to the feeder terminal unit, periodically sample the current transformer and voltage transformer of the distribution network in the power system according to the preset sampling period, and obtain the current-voltage three-phase data sequence.

[0031] Specifically, the feeder terminal unit is a device installed in the switch station, ring network cabinet or substation in the distribution automation system, used for collecting, monitoring, protecting and controlling the electrical parameters of the connected line. The feeder terminal unit can communicate with the main station system to realize remote measurement and control and fault judgment, and ensure the reliability and automation level of the distribution network operation. The feeder terminal unit performs data sampling according to the preset sampling period. The preset sampling period refers to the time interval set in advance by the person skilled in the art according to the actual situation, such as data collection every 5 milliseconds, 100 milliseconds or 1 second. The current transformer and voltage transformer are used to convert high-voltage and high-current signals into low-voltage and low-current standard signals in proportion, so that the subsequent equipment can be safely received.

[0032] Three-phase electricity refers to the three-phase line (generally marked as A, B, C) existing in the power system, which has the advantages of high transmission efficiency and smooth motor operation. By periodically sampling the current transformer and voltage transformer of the distribution network in the power system, the current-voltage three-phase data sequence is obtained, and the values of A, B and C three-phase current and three-phase voltage changing with time are obtained, and are organized and stored in the form of time sequence. For example, in the case of sampling 10 times per second, 10 seconds can form a sequence containing 60 groups (3-phase current x 10 times + 3-phase voltage x 10 times) of data.

[0033] S3: Based on the real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameters, and the current-voltage three-phase data sequence, state analysis is performed to trigger a fault warning instruction.

[0034] Specifically, the real-time gas concentration-generation rate is obtained by the oil chromatographic analyzer, which is used to monitor the concentration change rate of fault indication gases such as acetylene and carbon monoxide in the transformer oil. If the gas concentration-generation rate abnormally rises, it may indicate the presence of internal arc, overheating or insulation deterioration and other problems. The real-time temperature rise monitoring value is detected by the winding temperature sensor, which can reflect the temperature rise characteristics of the core components of the transformer under high load or abnormal state. If the real-time temperature rise monitoring value exceeds the design threshold, it may indicate that the heat dissipation system fails or the load is too heavy. The real-time integrated monitoring parameter is provided by the microcomputer monitoring device, which integrates multiple indicators such as three-phase current, voltage, frequency, load, arc signal, etc., reflecting the load fluctuation, voltage stability and phase balance of the transformer in the distribution network. For example, when the current-voltage data presents a sharp fluctuation or asymmetric trend, it may indicate that a sudden short circuit or load mutation occurs in the external line, and the power grid disturbance will be transmitted to the transformer body and cause abnormal internal parameters. When the real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameter, and the current-voltage three-phase data sequence are found to have a trend abnormality or sharp deviation through state analysis, a fault warning instruction will be triggered. The fault warning instruction will be uploaded to the SCADA system or local linkage device, and may drive subsequent actions such as alarm, interlock removal, dispatching instruction generation, etc., to ensure that the problem is responded to before it expands.

[0035] S4: performing feature backtracking on the perception module based on the fault warning instruction to generate a transformer feature backtracking result.

[0036] Specifically, feature backtracking is performed on the perception module based on the fault warning instruction. After detecting a potential fault and issuing a warning instruction, the perception module will start a historical data review and analysis process within a certain time range. By backtracking the data evolution trajectory within a certain time range, it is identified whether there are trend, mutation or periodic abnormal features, and a transformer feature backtracking result is generated.

[0037] S5: performing trend feature identification on the current-voltage three-phase data sequence to generate current-voltage three-phase data trend features.

[0038] Specifically, trend feature identification is performed on the current-voltage three-phase data sequence. The three-phase current and voltage data collected within a continuous time period are used as analysis data, and signal processing and analysis algorithms are applied to identify the trend features in the analysis data. By analyzing the evolution pattern of the data over time, indicators representing changes in operating state are extracted. In the identification process, such as moving average, local weighted regression, multi-scale wavelet analysis, Fourier transform, etc., can help to strip out transient interference and high-frequency noise, identify trend components, generate current-voltage three-phase data trend features, and provide key support for subsequent fault detection, anomaly identification and operation optimization models.

[0039] S6: performing fault early warning analysis according to the transformer feature backtracking result and the current-voltage three-phase data trend feature, and obtaining a fault early warning result.

[0040] Specifically, the fault early warning analysis is performed according to the transformer feature backtracking result and the current-voltage three-phase data trend feature. The transformer key state evolution features backtracked from the perception module are comprehensively analyzed with the dynamic trend features extracted from the current and voltage data, to identify potential abnormal evolution patterns, and then generate a fault early warning result with time foresight. The fault early warning analysis fuses the two types of information by using a rule engine, a machine learning model or an expert system, etc. For example, a warning rule is set: if the combustible gas generation rate in the transformer oil increases by more than 40 microliters per minute within 5 minutes, and the fluctuation of any one phase of the three-phase voltage exceeds 5%, a potential overheat fault warning is triggered. A neural network model can also be constructed to input the backtracking features and trend features, and output a warning risk score with a confidence level of 0 to 1. When the score is higher than 0.8, an automatic warning is issued.

[0041] Further, the application also includes that the perception module has a data buffer, and the data buffer is used for data storage of the feature backtracking window.

[0042] Specifically, the data buffer of the perception module is a memory area for short-term historical data storage, which temporarily caches recent data for quick extraction when analysis or backtracking is needed. The buffer mechanism ensures that in the scene of intensive data collection, the system can respond in time and will not cause processing efficiency to decrease due to data accumulation.

[0043] The data buffer is mainly set for the feature backtracking window. The feature backtracking window refers to a specific time range. Within the feature backtracking window, the system can backtracking analyze the running track and parameter evolution trend of the equipment, to determine whether the current anomaly has early signs or a gradual evolution process. For example, if a transformer triggers a warning at 13:40, the data from 13:30 to 13:40 is called to form a 10-minute backtracking window, from which continuous features such as gradual increase of temperature rise and slow growth of gas concentration are extracted, to provide evidence support for judging the fault trend.

[0044] Further, the application also includes: based on the fault early warning instruction, calling the backtracking data stored in the data buffer memory of the perception module, obtaining the gas concentration-generation rate backtracking data sequence, the temperature rise monitoring value backtracking data sequence and the integrated monitoring parameter backtracking data sequence; taking the real-time gas concentration-generation rate, the real-time temperature rise monitoring value and the real-time integrated monitoring parameter as foreground data, and taking the gas concentration-generation rate backtracking data sequence, the temperature rise monitoring value backtracking data sequence and the integrated monitoring parameter backtracking data sequence as background data; performing mapping feature backtracking analysis on the foreground data and the background data to generate a transformer feature backtracking result.

[0045] Specifically, when the fault early warning instruction is detected, the historical monitoring data in the past period of time, i.e., the backtracking data, is called from the data buffer of the perception module. The data buffer saves the continuous monitoring information of the transformer during the operation before the fault, for example, the gas concentration-generation rate refers to the concentration growth rate of the combustible gas in the transformer oil per unit time, reflecting the degree of insulation oil degradation or overheating; the temperature rise monitoring value represents the temperature change of the key parts of the transformer, such as the temperature rise of the winding relative to the environment, in units of degrees Celsius; the integrated monitoring parameter is a comprehensive index obtained by fusion, cleaning or weighted calculation of multiple original monitoring indexes, used to describe the overall health level or operation stability of the equipment. The backtracking data forms three time sequences, respectively recording the change process of gas concentration, temperature rise and comprehensive parameter.

[0046] Then, the monitoring data collected at the current time is defined as foreground data, representing the current state of the transformer, including the current gas generation rate, the current temperature rise and the current comprehensive state index. At the same time, the data in the period before the fault early warning, i.e., the backtracking data sequence obtained from the data buffer, is defined as background data, used to represent the historical operation state of the equipment before the early warning.

[0047] Subsequently, mapping feature backtracking analysis is performed on the foreground data and the background data. Mapping feature refers to one-to-one comparison and mathematical processing of the foreground data and the background data, for example, calculating the deviation amplitude, change rate, fluctuation trend, etc. of a certain index. Mapping feature backtracking analysis relies on feature extraction algorithm or pattern recognition model, which is used to mine the change relationship between different dimension parameters before and after the fault early warning, and then identify the feature trajectory related to the formation of the fault, for example, the gas concentration generation rate increases from 30 microliters per minute to 150 microliters per minute in the past 5 minutes, and the temperature rise value is stable above 80 degrees Celsius. The change pattern is likely to be an early feature caused by local overheating or insulation oil decomposition.

[0048] Further, the application also includes: extracting real-time gas concentration-generation rate in the foreground data, mapping deviation analysis with the gas concentration-generation rate backtracking data sequence in the background data, obtaining the gas concentration-generation rate deviation value sequence; center value screening of the gas concentration-generation rate deviation value sequence to determine the target backtracking center value; using the target backtracking center value to perform feature analysis data screening on the gas concentration-generation rate deviation value sequence, and performing trend feature analysis on the screening result to generate gas concentration-generation rate backtracking features; mapping feature backtracking analysis on the foreground data and the background data respectively to generate temperature rise monitoring value backtracking features and integrated monitoring parameter backtracking features, and combining the gas concentration-generation rate backtracking features to obtain the transformer feature backtracking result.

[0049] Specifically, the real-time gas concentration-generation rate is extracted from the foreground data, reflecting the growth rate of the gas concentration in the oil of the transformer per unit time, such as acetylene or hydrogen, to determine whether the transformer has a thermal fault or a discharge fault. At the same time, the gas concentration-generation rate sequence recorded in the background data in the previous period is extracted, reflecting the gas generation change trajectory of the equipment in the normal state. Then, the current real-time value is compared with the background historical sequence point by point, the deviation is calculated, and the gas concentration-generation rate deviation value sequence is formed to obtain the change amplitude of the current state compared with the historical state.

[0050] Next, the center value screening is performed on the gas concentration-generation rate deviation value sequence, the purpose of which is to find representative data points from the gas concentration-generation rate deviation value sequence, such as using the median or sliding mean of the statistical method to determine a stable reference value as the target backtracking center value.

[0051] Subsequently, the target backtracking center value is used to perform feature analysis data screening on the gas concentration-generation rate deviation value sequence, to exclude abnormal values with too small or too large deviations, and only the representative part is retained for further analysis. The screened values are used for trend feature analysis to observe whether they show increasing, decreasing, periodic change, etc. behavior patterns, thereby forming the gas concentration-generation rate backtracking features.

[0052] The same mapping feature backtracking analysis is performed on the temperature rise monitoring value and the integrated monitoring parameter, and the temperature rise monitoring value backtracking features and the integrated monitoring parameter backtracking features are generated by calculating the deviation value sequence, screening the center value, and performing feature analysis. The temperature rise monitoring value represents the real-time and historical temperature difference of the transformer, and the integrated monitoring parameter is a health parameter formed by fusing multiple basic indicators such as voltage, current, and load state through modeling method.

[0053] Finally, the gas concentration-generation rate backtracking feature, the temperature rise monitoring value backtracking feature, and the integrated monitoring parameter backtracking feature are fused to generate a transformer feature backtracking result, reflecting the state evolution process and main change trend of the transformer before the fault is formed, which is used for subsequent fault type discrimination, severity classification, and operation and maintenance strategy formulation. For example, if all three features show a continuous upward trend, it can be determined that the equipment is in a high-risk interval.

[0054] Further, the application further comprises: extracting the maximum deviation value in the gas concentration-generation rate deviation value sequence as an initial backtracking data center value; performing deep embedded center value iteration on the gas concentration-generation rate deviation value sequence based on the initial backtracking data center value to determine a target backtracking data center value.

[0055] Specifically, the maximum deviation value is extracted from the gas concentration-generation rate deviation value sequence. The gas concentration-generation rate deviation value represents the difference between the generation rate per unit time of a certain type of gas (such as acetylene, methane, hydrogen, etc.) at the current time and its historical reference value. The largest value in the entire set of deviation values is selected as the initial backtracking data center value, which is considered as the most significant manifestation point of potential faults, and is used to guide subsequent analysis.

[0056] Subsequently, based on the initial backtracking data center value, deep embedded center value iteration analysis is performed to estimate the initial backtracking data center value in the overall distribution and fluctuation pattern of the gas concentration-generation rate deviation value sequence. Instead of simply taking the average or median, the deviation value sequence is filtered and converged in multiple rounds by combining methods such as sliding window, outlier removal, and piecewise fitting. In each round, data points with smaller deviation from the current initial backtracking data center value are selected and retained, and a new center value is calculated again until the center value is stable. This can reduce the interference of extreme abnormal values on the judgment result, thereby more accurately determining the target backtracking data center value. For example, if the initial maximum deviation value is 180 microliters per minute, in the first iteration, data with a deviation value between 150 and 180 is retained, and the average is calculated to be 165; in the second iteration, data with a deviation value between 160 and 170 is further retained, and the final center value converges to about 165 microliters per minute, which more accurately reflects the core level of overall trend changes.

[0057] Further, the application further comprises: pre-constructing a trend feature recognition component, wherein the trend feature recognition component comprises a multi-scale trend feature recognition branch set and a multi-scale trend feature fusion branch; performing multi-scale analysis on the current-voltage three-phase data sequence using the multi-scale trend feature recognition branch set to determine a multi-scale trend feature set; and using the multi-scale trend feature fusion branch to interactively enhance the multi-scale trend feature set to obtain the current-voltage three-phase data trend feature.

[0058] Specifically, a trend feature recognition component is pre-constructed, which is used to identify and extract key features with trend or evolution rules in power system monitoring data. The trend feature recognition component internally includes a multi-scale trend feature recognition branch set and a multi-scale trend feature fusion branch. The multi-scale trend feature recognition branch set refers to a set of parallel analysis channels, each branch independently processes the input data based on different time scales (e.g. 1 minute, 5 minutes, 15 minutes, 1 hour, etc.) to capture the data trend at different time resolutions. The multi-scale trend feature fusion branch is used to integrate the multi-scale trend feature recognition branch set, which comprehensively analyzes the trend features at multiple scales through comparison, cooperation and enhancement mechanisms to avoid one-sidedness of trend judgment at a single scale and improve the accuracy and stability of trend recognition. The fusion process can be achieved through feature splicing, attention mechanism or weighted average strategy, and finally outputs a trend feature result that fully describes the data evolution characteristics.

[0059] Then, the multi-scale trend feature recognition branch set is used to process the current-voltage three-phase data sequence. The current-voltage three-phase data sequence refers to the synchronous data stream collected from the three-phase power grid (A, B, C three-phase), which contains the specific values of current and voltage of each phase changing over time. Through analysis of different time windows and filtering strategies, various trend features such as load increase, voltage abnormal fluctuation or frequent switching can be identified, forming a multi-scale trend feature set.

[0060] Then, the multi-scale trend feature fusion branch is used to interactively enhance the extracted multi-scale trend feature set. Interactive enhancement means that the features extracted by different branches are not used in isolation, but interact through correlation calculation, feature alignment, attention mechanism, etc. to extract higher-level association patterns, generate complete current-voltage three-phase data trend features, and serve as key inputs for judging the health status of the power distribution system and predicting potential abnormalities.

[0061] Further, the application also includes: calling a feature similarity identification component in the multi-scale trend feature fusion branch; randomly extracting two multi-scale trend features from the multi-scale trend feature set without replacement, using the feature similarity identification component to identify the similarity of the two multi-scale trend features, and obtaining a feature similarity set; calling an interaction matrix construction component in the multi-scale trend feature fusion branch; using the interaction matrix construction component to normalize the feature similarity set, and filling the processed result into an initially empty matrix to obtain an interaction matrix; using the interaction matrix to perform feature enhancement on the two multi-scale trend features respectively, generating two enhanced multi-scale trend features; and iteratively obtaining a multi-scale enhanced trend feature set, and performing mean value processing on the multi-scale enhanced trend feature set to obtain a current-voltage three-phase data trend feature.

[0062] Specifically, the feature similarity identification component in the multi-scale trend feature fusion branch is called to start a sub-module specially used for analyzing the similarity between trend features. The feature similarity identification component is a tool for calculating the similarity between two data features in space or numerical pattern, and uses algorithms such as Euclidean distance, cosine similarity or dynamic time warping to evaluate whether the trends at different time scales are consistent.

[0063] Next, two multi-scale trend features are randomly extracted from the multi-scale trend feature set without replacement, that is, trend data segments are selected from the entire multi-scale trend library without repeated extraction. The two extracted multi-scale trend features are identified for similarity by the feature similarity identification component, which can determine whether the trends of the two multi-scale trend features have common patterns or evolution paths, and obtain a feature similarity set.

[0064] Then, the interaction matrix construction component in the multi-scale trend feature fusion branch is called. The interaction matrix construction component standardizes (i.e., normalizes) the feature similarity set so that it falls between 0 and 1, and sequentially fills the standard values into an originally empty two-dimensional matrix. The rows and columns of the interaction matrix correspond to the compared trend features, and each element in the interaction matrix represents the similarity score of the corresponding feature pair.

[0065] Next, the interaction matrix is used to enhance the features of the two compared multi-scale trend features. Feature enhancement refers to weighting correction, bias adjustment or context expansion of each multi-scale trend feature according to the similar structures between the multi-scale trend features identified in the interaction matrix, so as to improve its representation ability and make the multi-scale trend feature more comprehensively reflect the system state change.

[0066] According to the same method, the above process is repeatedly performed to generate a set composed of multiple enhanced multi-scale enhanced trend features, referred to as a multi-scale enhanced trend feature set. Finally, all multi-scale enhanced trend features in the multi-scale enhanced trend feature set are subjected to mean value processing, that is, all enhanced multi-scale trend features are averaged item by item, so as to obtain current-voltage three-phase data trend features for state prediction, fault trend judgment and other tasks.

[0067] Further, the application further comprises: determining a warning feedback verification window based on the fault warning result; and performing fault troubleshooting verification on the transformer and the power distribution network in the warning feedback verification window to obtain a verification result.

[0068] Specifically, the warning feedback verification window is determined based on the fault warning result. After identifying the potential fault risk or abnormal trend, according to the time point and risk level of the warning event, a time period is set as the key observation period for subsequent verification, referred to as the warning feedback verification window, which can be 5 minutes, 15 minutes or longer. The specific length is determined by the severity of the risk event, the device response time and the system tolerance. The purpose of the warning feedback verification window is to closely observe the device operating state before and after the possible fault occurs to confirm whether the warning is accurate.

[0069] In the warning feedback verification window, the transformer and the power distribution network are subjected to fault troubleshooting verification. Through on-site inspection, secondary sampling, device diagnosis algorithm or manual inspection, the object pointed to by the warning is carefully investigated. For example, the analysis intensity of parameters such as oil gas change trend, winding temperature fluctuation and load current anomaly can be enhanced within the warning feedback verification window, and historical data is compared to verify whether the previous warning truly reflects the abnormal state. The verification process is not limited to the transformer itself, but also includes the connection relationship in the power distribution network, the state of adjacent devices, and even the operation of auxiliary devices such as cables and circuit breakers. Finally, the verification result is obtained to determine whether the warning is an effective warning, a false alarm or a missed alarm.

[0070] Further, the application further comprises: based on the verification result, adjusting the preset sampling period of the feeder terminal unit and the feature backtracking window of the perception module.

[0071] Specifically, the preset sampling period of the feeder terminal unit is adjusted based on the verification result. After the verification link after the fault early warning is completed, it is judged whether the sampling frequency of the feeder terminal unit needs to be adjusted according to the verification result. The feeder terminal unit is responsible for collecting three-phase sequence data of current and voltage in the power distribution network, and its sampling period determines the time interval of data update. If the verification result shows that there is a false alarm or a false alarm, the current sampling period of 5 minutes may be shortened to 1 minute to improve the timeliness of data acquisition; on the contrary, in the case of long-term stability and no abnormality, the period may be extended to reduce resource consumption.

[0072] At the same time, the feature backtracking window of the perception module is adjusted based on the verification result. If the verification shows that part of the abnormal trend is not fully captured, it means that the feature backtracking window is too short, which may be expanded to 45 minutes or even 60 minutes; if there is too much response redundancy or significant noise interference, the feature backtracking window range can be reduced to improve response efficiency.

[0073] In summary, the fault early warning method for power transmission and distribution equipment provided by the present application has the following technical effects: by achieving the technical target of dynamic fault early warning analysis based on multi-source monitoring data fusion and trend feature evolution analysis, the technical effects of improving early warning accuracy, enhancing system adaptability, reducing false alarm rate and false alarm rate, realizing early fault identification and auxiliary decision support are achieved.

[0074] Embodiment two, based on the same inventive concept as the fault early warning method for power transmission and distribution equipment in the foregoing embodiments, the present application also provides a fault early warning system for power transmission and distribution equipment, please refer to the attached Figure 2, comprising: a perception module construction module 11 for constructing a perception module for health state monitoring of a transformer of a power system using an oil chromatographic analyzer, a winding temperature sensor, and a microcomputer monitoring device, obtaining real-time gas concentration-generation rate, real-time temperature rise monitoring value, and real-time integrated monitoring parameters of the transformer through real-time monitoring of the transformer by the perception module; a periodic sampling module 12 for periodically sampling current transformers and voltage transformers of a distribution network in the power system according to a preset sampling period of a feeder terminal unit, obtaining current-voltage three-phase data sequences; a state analysis module 13 for performing state analysis based on the real-time gas concentration-generation rate, the real-time temperature rise monitoring value, and the real-time integrated monitoring parameters, and the current-voltage three-phase data sequences, triggering a fault warning instruction; a feature backtracking module 14 for performing feature backtracking on the perception module based on the fault warning instruction, generating a transformer feature backtracking result; a trend feature identification module 15 for performing trend feature identification on the current-voltage three-phase data sequences, generating current-voltage three-phase data trend features; and a fault warning analysis module 16 for performing fault warning analysis according to the transformer feature backtracking result and the current-voltage three-phase data trend features, obtaining a fault warning result.

[0075] Further, the fault warning system for power transmission, transformation, and distribution equipment is further used for: the perception module having a data buffer area, the data buffer area being used for data storage of a feature backtracking window.

[0076] Further, the fault warning system for power transmission, transformation, and distribution equipment is further used for: based on the fault warning instruction, retrieving backtracking data stored in the data buffer area of the perception module, obtaining a gas concentration-generation rate backtracking data sequence, a temperature rise monitoring value backtracking data sequence, and an integrated monitoring parameter backtracking data sequence; taking the real-time gas concentration-generation rate, the real-time temperature rise monitoring value, and the real-time integrated monitoring parameters as foreground data, and taking the gas concentration-generation rate backtracking data sequence, the temperature rise monitoring value backtracking data sequence, and the integrated monitoring parameter backtracking data sequence as background data; performing mapping feature backtracking analysis on the foreground data and the background data, and generating a transformer feature backtracking result.

[0077] Further, the fault early warning system for power transmission and distribution equipment is also used for: extracting the real-time gas concentration-generation rate in the foreground data, mapping and deviation analysis with the gas concentration-generation rate backtracking data sequence in the background data, obtaining a gas concentration-generation rate deviation value sequence; center value screening is performed on the gas concentration-generation rate deviation value sequence to determine a target backtracking center value; the target backtracking center value is used for feature analysis data screening on the gas concentration-generation rate deviation value sequence, and trend feature analysis is performed on the screening result to generate a gas concentration-generation rate backtracking feature; mapping feature backtracking analysis is performed on the foreground data and the background data respectively to generate a temperature rise monitoring value backtracking feature and an integrated monitoring parameter backtracking feature, and the transformer feature backtracking result is obtained in combination with the gas concentration-generation rate backtracking feature.

[0078] Further, the fault early warning system for power transmission and distribution equipment is also used for: extracting the maximum deviation value in the gas concentration-generation rate deviation value sequence as an initial backtracking data center value; performing deep embedded center value iteration on the gas concentration-generation rate deviation value sequence based on the initial backtracking data center value to determine a target backtracking data center value.

[0079] Further, the fault early warning system for power transmission and distribution equipment is also used for: pre-constructing a trend feature recognition component, wherein the trend feature recognition component includes a multi-scale trend feature recognition branch set and a multi-scale trend feature fusion branch; performing multi-scale analysis on the current-voltage three-phase data sequence by using the multi-scale trend feature recognition branch set to determine a multi-scale trend feature set; performing interactive enhancement on the multi-scale trend feature set by using the multi-scale trend feature fusion branch to obtain the current-voltage three-phase data trend feature.

[0080] Further, the fault early warning system for power transmission and distribution equipment is also used for: calling a feature similarity recognition component in the multi-scale trend feature fusion branch; extracting two multi-scale trend features from the multi-scale trend feature set without replacement, performing similarity recognition on the two multi-scale trend features by using the feature similarity recognition component to obtain a feature similarity set; calling an interaction matrix construction component in the multi-scale trend feature fusion branch; performing normalization processing on the feature similarity set by using the interaction matrix construction component, and filling the processed result into an initially empty matrix to obtain an interaction matrix; performing feature enhancement on the two multi-scale trend features respectively by using the interaction matrix to generate two enhanced multi-scale trend features; and in this way, a multi-scale enhanced trend feature set is obtained, and mean value processing is performed on the multi-scale enhanced trend feature set to obtain the current-voltage three-phase data trend feature.

[0081] Further, the fault early warning system for power transmission, transformation and distribution equipment is further used for determining a warning feedback verification window based on the fault early warning result; and performing fault troubleshooting verification on the transformer and the power distribution network in the warning feedback verification window to obtain a verification result.

[0082] Further, the fault early warning system for power transmission, transformation and distribution equipment is further used for determining a warning feedback verification window based on the fault early warning result; and performing fault troubleshooting verification on the transformer and the power distribution network in the warning feedback verification window to obtain a verification result.

[0083] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The fault early warning method for power transmission, transformation and distribution equipment in the aforementioned embodiment one is also applicable to the fault early warning system for power transmission, transformation and distribution equipment in the present embodiment. Through the foregoing detailed description of the fault early warning method for power transmission, transformation and distribution equipment, those skilled in the art can clearly know the fault early warning system for power transmission, transformation and distribution equipment in the present embodiment. Therefore, for the sake of brevity of the specification, the fault early warning system for power transmission, transformation and distribution equipment in the present embodiment will not be described in detail.

[0084] The above description of disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0085] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application also intends to include these modifications and variations.

Claims

1. A method for failure pre-warning of power transmission and distribution equipment, characterized in that, The method comprises: A sensing module for health state monitoring of a transformer of a power system is constructed by using an oil chromatographic analyzer, a winding temperature sensor and a microcomputer monitoring device, real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameters of the transformer are obtained by real-time monitoring of the transformer through the sensing module, and the real-time integrated monitoring parameters include three-phase voltage, current, active / reactive power, frequency and load current; According to the feeder terminal unit, the current and voltage transformers of the distribution network in the power system are periodically sampled according to a preset sampling period, and a current-voltage three-phase data sequence is obtained; Based on the real-time gas concentration-generation rate, the real-time temperature rise monitoring value and the real-time integrated monitoring parameters, and the current-voltage three-phase data sequence, state analysis is performed to trigger a fault warning instruction; Based on the fault warning instruction, feature backtracking of the sensing module is performed to generate a transformer feature backtracking result; Trend feature identification is performed on the current-voltage three-phase data sequence to generate current-voltage three-phase data trend features; According to the transformer feature backtracking result and the current-voltage three-phase data trend features, fault warning analysis is performed to obtain a fault warning result; The sensing module has a data buffer area, and the data buffer area is used for data storage of a feature backtracking window; Based on the fault warning instruction, feature backtracking of the sensing module is performed to generate a transformer feature backtracking result, including: Based on the fault warning instruction, the backtracking data stored in the data buffer area of the sensing module is retrieved to obtain a gas concentration-generation rate backtracking data sequence, a temperature rise monitoring value backtracking data sequence and an integrated monitoring parameter backtracking data sequence; The real-time gas concentration-generation rate, the real-time temperature rise monitoring value and the real-time integrated monitoring parameters are taken as foreground data, and the gas concentration-generation rate backtracking data sequence, the temperature rise monitoring value backtracking data sequence and the integrated monitoring parameter backtracking data sequence are taken as background data; Mapping feature backtracking analysis is performed on the foreground data and the background data to generate a transformer feature backtracking result; Trend feature identification is performed on the current-voltage three-phase data sequence to generate current-voltage three-phase data trend features, including: A trend feature identification component is pre-constructed, wherein the trend feature identification component includes a multi-scale trend feature identification branch set and a multi-scale trend feature fusion branch; The multi-scale trend feature identification branch set is used to perform multi-scale analysis on the current-voltage three-phase data sequence to determine a multi-scale trend feature set; The multi-scale trend feature fusion branch is used to perform interactive enhancement on the multi-scale trend feature set to obtain the current-voltage three-phase data trend features.

2. The method for failure early warning of power distribution equipment according to claim 1, wherein, Including: The real-time gas concentration-generation rate in the foreground data is extracted and mapping deviation analysis is performed on the gas concentration-generation rate backtracking data sequence in the background data to obtain a gas concentration-generation rate deviation value sequence; Center value screening is performed on the gas concentration-generation rate deviation value sequence to determine a target backtracking center value; Perform feature analysis data screening on the gas concentration-generation rate deviation value sequence using the target backtracking center value, and perform trend feature analysis on the screening result to generate a gas concentration-generation rate backtracking feature; Perform mapping feature backtracking analysis on the foreground data and the background data respectively to generate a temperature rise monitoring value backtracking feature and an integrated monitoring parameter backtracking feature, and combine the gas concentration-generation rate backtracking feature to obtain the transformer feature backtracking result.

3. The method for failure early warning of power distribution equipment according to claim 1, wherein, Comprise: Extract the maximum deviation value in the gas concentration-generation rate deviation value sequence as the initial backtracking data center value; Based on the initial backtracking data center value, the gas concentration-generation rate deviation value sequence is deeply embedded and iterated to determine the target backtracking data center value.

4. The method for failure early warning of power distribution equipment according to claim 1, wherein, Utilize the multi-scale trend feature fusion branch to interactively enhance the multi-scale trend feature set to obtain the current-voltage three-phase data trend feature, comprising: Call the feature similarity recognition component in the multi-scale trend feature fusion branch; Two multi-scale trend features are randomly extracted from the multi-scale trend feature set without replacement, and the feature similarity recognition component is used to identify the similarity of the two multi-scale trend features to obtain a feature similarity set; Call the interaction matrix construction component in the multi-scale trend feature fusion branch; Use the interaction matrix construction component to normalize the feature similarity set and fill the processed result into the initially empty matrix to obtain an interaction matrix; Use the interaction matrix to perform feature enhancement on the two multi-scale trend features respectively to generate two enhanced multi-scale trend features; By analogy, a multi-scale enhanced trend feature set is obtained, and the multi-scale enhanced trend feature set is processed by mean value to obtain the current-voltage three-phase data trend feature.

5. The method for failure pre-warning of power distribution equipment according to claim 1, wherein, Comprise: Determine a warning feedback verification window based on the fault warning result; Verify the transformer and the distribution network for faults in the warning feedback verification window to obtain a verification result.

6. The method for failure pre-warning of power distribution equipment according to claim 5, wherein, Based on the verification result, the preset sampling period of the feeder terminal unit and the feature backtracking window of the perception module are adjusted.

7. A fault warning system for power distribution equipment, characterized by, For implementing the steps of the fault warning method for power transmission and distribution equipment in any one of claims 1-6, comprising: A perception module construction module is used to construct a perception module for health state monitoring of the transformer of the power system by using an oil chromatographic analyzer, a winding temperature sensor and a microcomputer monitoring device, and to obtain real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameter by real-time monitoring of the transformer through the perception module; A periodic sampling module is used to periodically sample the current transformer and the voltage transformer of the distribution network in the power system according to the preset sampling period of the feeder terminal unit to obtain a current-voltage three-phase data sequence; A state analysis module is used to perform state analysis based on the real-time gas concentration-generation rate, real-time temperature rise monitoring value and real-time integrated monitoring parameter, and the current-voltage three-phase data sequence to trigger a fault warning instruction; a feature backtracking module, configured to perform feature backtracking on the perception module based on the fault early warning instruction, to generate a transformer feature backtracking result; a trend feature identification module, configured to perform trend feature identification on the current-voltage three-phase data sequence, to generate a current-voltage three-phase data trend feature; a fault early warning analysis module, configured to perform fault early warning analysis according to the transformer feature backtracking result and the current-voltage three-phase data trend feature, to obtain a fault early warning result.

Images