A transformer comprehensive monitoring and early warning method and system

Abstract

The application discloses a transformer comprehensive monitoring and early warning method and system, and relates to the field of transformer monitoring.The method comprises the following steps: collecting state data of a target transformer in real time, wherein the state data comprises oil chromatogram data, partial discharge signals, grounding currents, local temperature data and local vibration signals; pre-processing the current state data; performing multi-angle evaluation on the state of the target transformer according to the pre-processed current state data; and performing early warning on each abnormal state of the target transformer if the state of the target transformer is abnormal according to at least one angle evaluation.The application realizes accurate monitoring and early warning of the state of the target transformer, and solves the problems of low monitoring precision and insufficient fault early warning capability of the existing transformer monitoring method.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of transformer monitoring, in particular to a transformer comprehensive monitoring and early warning method and system. BACKGROUND

[0002] Power transformer operation health state monitoring is one of the key links to ensure the stable operation of the power system. By monitoring the operation state of the transformer, potential faults can be detected in time, preventive measures can be taken, and power outages and other economic losses caused by equipment failure can be avoided. In the prior art, when a transformer has a suspected fault or defect but the specific nature of the fault has not been determined, certain monitoring equipment and measures are used to allow the transformer to continue operating for a period of time to collect data for further evaluation of the impact of the fault on the performance of the transformer and whether it poses a threat to the safety of the power grid.

[0003] However, the conventional transformer monitoring method uses a single or a few sensors to collect single state data of the transformer and compares it with a preset threshold, and then performs early warning according to the comparison result, or uses a few state data to perform weighted summation and compares the weighted summation result with a preset threshold, and then performs early warning according to the comparison result, to realize the state monitoring of the transformer. The monitoring precision of this transformer monitoring method is low and the fault early warning capability is insufficient. SUMMARY

[0004] The purpose of the present application is to provide a transformer comprehensive monitoring and early warning method and system to solve the problem of low monitoring precision and insufficient fault early warning capability of the existing transformer monitoring method.

[0005] To achieve the above-mentioned purpose, the present application provides the following solutions:

[0006] In a first aspect, the present application provides a transformer comprehensive monitoring and early warning method, comprising:

[0007] real-time collection of state data of a target transformer, the state data including oil chromatogram data, partial discharge signal, ground current, local temperature data and local vibration signal;

[0008] preprocessing of the current state data;

[0009] multi-angle evaluation of the state of the target transformer according to the preprocessed current state data;

[0010] if at least one angle evaluation obtains an abnormal state of the target transformer, early warning of each abnormal state of the target transformer.

[0011] Optionally, before the multi-angle evaluation of the state of the target transformer according to the preprocessed current state data, the method further comprises:

[0012] a first machine learning model is constructed, and the first machine learning model is parameter-optimized by using a particle swarm algorithm and InputData1 and T1, wherein InputData1 is a plurality of local temperature data of a sample transformer, T1 is an overall temperature of the sample transformer, InputData1 is input of the first machine learning model, T1 is output of the first machine learning model, and an optimal parameter combination is found through iterative search:

[0013] θ opt = argmin θ Error(f ML (InputData1, θ j ));

[0014] wherein f ML represents the first machine learning model, θ j represents a first influencing factor, wherein j = {1, 2, …, n1}, n1 represents a quantity of the first influencing factor, the first influencing factor refers to a factor affecting T1, Error represents a prediction error function of the first machine learning model, argmin θ Error represents a parameter combination of the first machine learning model when a prediction error Error of the first machine learning model is minimum, and θ opt represents an optimal parameter combination found by the particle swarm algorithm.

[0015] The multi-angle evaluation of the state of the target transformer according to the preprocessed current state data comprises:

[0016] a current overall temperature T2 of the target transformer is predicted by using the first machine learning model with a parameter combination θ opt according to a plurality of second influencing factors θ i and current InputData2, wherein i = {1, 2, …, n2}, n2 represents a quantity of the second influencing factor, the second influencing factor refers to a factor affecting T2, and InputData2 represents a plurality of local temperature data of the target transformer.

[0017] a temperature rise rate Δv T of the target transformer in Δt time is calculated:

[0018]

[0019] Where T'2 represents the overall temperature of the target transformer predicted in the previous test, Δt = t2 - t1, t2 represents the current data collection time of InputData2, and t1 represents the previous data collection time of InputData2.

[0020] If Δv T A temperature rise exceeding a preset threshold indicates an abnormal temperature rise in the target transformer.

[0021] Optionally, the step of evaluating the state of the target transformer from multiple perspectives based on the preprocessed current state data further includes:

[0022] According to multiple second influencing factors θ i Combine the current InputData2 with parameters to form θ opt After the first machine learning model predicts the current overall temperature T2 of the target transformer, it then considers multiple second influencing factors θ. i Based on the Bayesian model and the current InputData2, P(T2>T) is determined according to the following formula. threshold ,θ i |InputData2):

[0023]

[0024] Wherein, P(InputData2|T2>T) threshold ,θ i ) indicates that it is greater than T threshold T2 and multiple secondary influencing factors θ i The likelihood of InputData2 is observed below, P(T2>T). threshold ,θ i ) indicates greater than T threshold T2 and multiple secondary influencing factors θ i The prior probability, P(InputData2), represents the marginal likelihood of InputData2, P(T2>T) threshold ,θ i |InputData2) indicates that it is greater than T threshold T2 and multiple secondary influencing factors θ i The posterior probability; where T threshold The set temperature threshold;

[0025] If P(T2>T) threshold ,θ i If InputData2) is greater than the set probability threshold, it indicates that the temperature of the target transformer is abnormal.

[0026] Optionally, the current state data is preprocessed, specifically including:

[0027] The single discharge pulse of the current partial discharge signal is subjected to time domain analysis, frequency domain analysis and time-frequency analysis, and time domain features, frequency domain features and time-frequency features of the partial discharge signal are extracted;

[0028] The state of the target transformer is evaluated from the preprocessed current state data, including:

[0029] The extracted time domain features, frequency domain features and time-frequency features of the partial discharge signal are input into the constructed insulation fault assessment model, and the insulation threat level corresponding to the insulation fault assessment model is output;

[0030] If the insulation threat level output by the insulation fault assessment model is greater than a preset insulation threat level threshold, it indicates that the insulation performance of the target transformer is abnormal.

[0031] Optionally, the state of the target transformer is evaluated from the preprocessed current state data, including:

[0032] If the frequency of at least part of the local vibration signals in the current multiple local vibration signals is an integer multiple of 50Hz and the amplitude is greater than a preset first amplitude threshold, it indicates that the winding of the target transformer is loose;

[0033] If the frequency of at least part of the local vibration signals in the current multiple local vibration signals is greater than 20Hz and less than 50Hz and the amplitude is greater than a preset second amplitude threshold, it indicates that the core of the target transformer is loose;

[0034] According to the grounding current, a first fault recognition model based on attention convolutional neural network or reinforcement learning is used to recognize the grounding fault and insulation damage fault of the target transformer.

[0035] Optionally, the state of the target transformer is evaluated from the preprocessed current state data, including:

[0036] According to the current oil chromatographic data, a second machine learning model is used to identify the overheat fault or discharge fault and fault position of the target transformer, wherein the second machine learning model includes a GA-SVM model or a CEABC-WNN model; wherein the GA-SVM model refers to a final SVM obtained by optimizing the parameter combination of a support vector machine SVM by a genetic algorithm GA and training; and the CEABC-WNN model refers to a final WNN model obtained by optimizing the initial parameter combination of a wavelet neural network WNN model by a chaos enhanced artificial bee colony algorithm CEABC and training.

[0037] Optionally, the multi-angle evaluation of the state of the target transformer according to the preprocessed current state data comprises:

[0038] calculating correlation coefficients between the preprocessed current state data;

[0039] constructing a correlation coefficient matrix according to the correlation coefficients between the state data;

[0040] constructing an undirected graph network based on the correlation coefficient matrix, wherein the name of the node of the undirected graph network is the name of the state data, the attribute value of the node is the corresponding state data value, and the edge is the connection line between the pair of significantly correlated state data; wherein if the correlation coefficients of two state data are greater than a preset correlation coefficient threshold, they are defined as a group of the pair of significantly correlated state data;

[0041] identifying independent communities in the undirected graph network through a community detection algorithm;

[0042] identifying an abnormal state of the target transformer according to the identified independent communities.

[0043] Optionally, the multi-angle evaluation of the state of the target transformer according to the preprocessed current state data comprises:

[0044] adopting a hierarchical clustering model based on a bottom-up aggregation strategy to cluster the preprocessed current state data;

[0045] performing comprehensive fault diagnosis according to the clustering result to identify a fault of the target transformer, the fault comprising at least one of a partial discharge fault, a mechanical fault, a grounding fault, an insulation fault and an overheat fault.

[0046] Optionally, the transformer comprehensive monitoring and early warning method further comprises:

[0047] constructing a digital twin model of the target transformer;

[0048] synchronously simulating the operation process of the target transformer through the digital twin model and collecting state data of the digital twin model in real time;

[0049] inputting an input sequence into a time series difference learning model to predict future state data of the target transformer through the time series difference learning model; wherein the input sequence comprises a plurality of pairs of input data, and each pair of input data comprises state data of the digital twin model and the target transformer at the same time;

[0050] The future state data of the target transformer is taken as the current state data of the target transformer, and the current state data is preprocessed and subsequent steps are performed to prewarn each abnormal state of the target transformer in the future.

[0051] In a second aspect, the application provides a transformer comprehensive monitoring and prewarning system, comprising:

[0052] A collection module is configured to collect real-time current state data of the target transformer, wherein the state data comprises oil chromatographic data, partial discharge signals, ground current, local temperature data and local vibration signals.

[0053] A data processing module is configured to pre-process the current state data.

[0054] A multi-angle evaluation module is configured to evaluate the state of the target transformer from multiple angles according to the pre-processed current state data.

[0055] A prewarning module is configured to prewarn each abnormal state of the target transformer if at least one angle evaluation indicates that the state of the target transformer is abnormal.

[0056] According to the embodiments of the application, the following technical effects are achieved:

[0057] The application provides a transformer comprehensive monitoring and early warning method and system, which acquires current state data of a target transformer in real time, the state data including oil chromatographic data, partial discharge signals, grounding currents, local temperature data and local vibration signals, compared with the prior art which only uses a single index or a few indexes for transformer state monitoring, the adopted indexes cover a wider range, avoids the blind area of using a single or a few indexes for transformer current state monitoring, and can more accurately perform transformer current state monitoring and abnormal early warning; the quality of the state data is improved through preprocessing of the current state data, and the accuracy and reliability of multi-angle evaluation of the state of the target transformer are improved; compared with the prior art, multi-angle evaluation can improve the accuracy of the current state evaluation of the transformer; when the state of the target transformer is abnormal in at least one angle evaluation, the target transformer is early warned for each abnormal state, and since the accuracy of the state evaluation of the target transformer is improved, the early warning capability for the abnormal state (fault) of the target transformer is improved; in summary, the application uses the preprocessed current oil chromatographic data, partial discharge signals, grounding currents, local temperature data and local vibration signals of the target transformer to perform multi-angle evaluation of the state of the target transformer, and when the state of the target transformer is abnormal in at least one angle evaluation, the target transformer is early warned for each abnormal state, thereby realizing accurate monitoring and early warning of the state of the target transformer, and solving the problems of low monitoring precision and insufficient fault early warning capability of the existing transformer monitoring method. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0059] Figure 1 A flowchart of a transformer comprehensive monitoring and early warning method provided by an embodiment of the present application;

[0060] Figure 2 A functional module diagram of a transformer comprehensive monitoring and early warning system provided by an embodiment of the present application;

[0061] Figure 3 A structural diagram of a computer device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0062] With reference to the drawings and embodiments, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0063] The above-mentioned purposes, features and advantages of the present application will be more apparent and understandable. The present application will be further described in detail below with reference to the drawings and specific embodiments.

[0064] In an exemplary embodiment, as shown in Figure 1 A transformer comprehensive monitoring and early warning method is provided, including the following steps 101 to 104. Among them:

[0065] Step 101, real-time acquisition of state data of the target transformer, the state data including oil chromatographic data, partial discharge signal, ground current, local temperature data and local vibration signal.

[0066] In the embodiments of the present application, the sensor used to collect the partial discharge signal is not specifically limited and can be selected as needed. For example, an oil valve type UHF high-frequency sensor is used to collect the partial discharge signal of the target transformer. The temperature includes the oil temperature and the shell temperature. The sensor used to collect the oil temperature and the shell temperature is not specifically limited and can be selected as needed. For example, a fiber Bragg grating temperature sensor arranged in the insulating oil tank of the target transformer is used to collect the oil temperature of the target transformer. For example, a PT100 temperature sensor arranged outside the insulating oil tank of the target transformer is used to collect the shell temperature of the target transformer. The sensor used to collect the ground current of the core of the target transformer is not specifically limited and can be selected as needed. For example, a high-frequency current transformer is used to collect the ground current of the core of the target transformer, and the high-frequency current transformer can use an open-close type core sensor. A vibration sensor is used to collect the local vibration signal of the target transformer. The oil chromatographic data refers to the gas component and the concentration data of each gas in the insulating oil of the target transformer. Through the oil valve on the body of the target transformer, the insulating oil of the target transformer is introduced into the degassing machine by one-in-one-out, which can ensure the circulation of the insulating oil under normal working conditions of the target transformer. The degassing machine uses a headspace degassing method to separate the gas in the insulating oil when the gas-liquid of the insulating oil reaches equilibrium. The mixed gas separated is separated into each single component gas by a chromatographic column, and then the concentration data of each gas is obtained through a gas-sensitive sensor. In the embodiments of the present application, H2, CH4, CO, C2H6, CO2, C2H4, C2H2, hydrocarbon gas and other components in the transformer insulating oil are detected in real time.

[0067] Step 102, preprocessing of the current state data.

[0068] In step 103, the state of the target transformer is evaluated from multiple angles according to the preprocessed current state data.

[0069] In step 104, if at least one angle evaluation shows that the state of the target transformer is abnormal, a warning is given for each abnormal state of the target transformer.

[0070] The steps 101 to 104 are implemented as follows: by collecting the current state data of the target transformer in real time, the state data including oil chromatographic data, partial discharge signals, ground current, local temperature data and local vibration signals, compared with the prior art which only uses a single indicator or a few indicators to monitor the state of the transformer, the indicators used cover a wider range, avoiding the blind area of using a single or a few indicators to monitor the current state of the transformer, and the current state of the transformer can be more accurately monitored and abnormal warning can be given; by preprocessing the current state data, the quality of the state data is improved, and the accuracy and reliability of the multi-angle evaluation of the state of the target transformer are improved; by evaluating the state of the target transformer from multiple angles according to the preprocessed current state data, compared with the prior art, the multi-angle evaluation can improve the accuracy of the current state evaluation of the transformer; by giving a warning for each abnormal state of the target transformer when at least one angle evaluation shows that the state of the target transformer is abnormal, since the accuracy of the state evaluation of the target transformer is improved, the warning ability for the abnormal state (fault) of the target transformer is also improved; in summary, by using the preprocessed current oil chromatographic data, partial discharge signals, ground current, local temperature data and local vibration signals of the target transformer, the state of the target transformer is evaluated from multiple angles, and a warning is given for each abnormal state of the target transformer when at least one angle evaluation shows that the state of the target transformer is abnormal, the accurate monitoring and warning of the state of the target transformer are realized, and the problem of low monitoring accuracy and insufficient fault warning ability of the existing transformer monitoring method is solved.

[0071] It should be noted that in the embodiments of the present application, the state of the target transformer is evaluated from multiple angles according to the preprocessed current state data, which includes evaluating the state of the target transformer according to the preprocessed current oil chromatographic data, partial discharge signals, ground current, local temperature data and local vibration signals respectively, and then obtaining the multi-angle evaluation result of the state of the target transformer. The evaluation of the state of the target transformer according to the preprocessed current oil chromatographic data, partial discharge signals, ground current, local temperature data and local vibration signals respectively will be described in detail below.

[0072] Before evaluating the state of the target transformer according to the preprocessed current local temperature data, the method further comprises:

[0073] A first machine learning model is constructed, and a particle swarm algorithm and InputData1 and T1 are used to optimize parameters of the first machine learning model, wherein InputData1 is a plurality of local temperature data of a sample transformer, T1 is an overall temperature of the sample transformer, InputData1 is used as an input of the constructed first machine learning model, T1 is used as an output of the constructed first machine learning model, and an optimal parameter combination is found through iterative search:

[0074] θ opt =argmin θ Error(f ML (InputData1,θ j ));

[0075] wherein f ML represents the first machine learning model, θ j represents a first influencing factor, wherein j={1,2,…,n1}, n1 represents a quantity of the first influencing factor, the first influencing factor refers to a factor affecting T1, Error represents a prediction error function of the first machine learning model, and argmin θ Error represents a parameter combination of the first machine learning model when a prediction error Error of the first machine learning model is minimum, and θ opt represents the optimal parameter combination found by the particle swarm algorithm.

[0076] In the embodiments of the application, the first machine learning model is preferably a neural network model, the parameters of the first machine learning model are optimized by using a PSO (particle swarm) algorithm, the optimal parameter combination is found through iterative search, and the prediction accuracy of the first machine learning model is effectively improved. The overall temperature of the sample transformer can be estimated according to past experience historical operation data through an empirical formula.

[0077] The state of the target transformer is evaluated according to the preprocessed current local temperature data, and the evaluation comprises the following steps 201 to 203. Wherein

[0078] In step 201, the current overall temperature T2 of the target transformer is predicted by using the first machine learning model with the parameter combination θ i and the current InputData2 according to a plurality of second influencing factors θ opt , wherein i={1,2,…,n2}, n2 represents a quantity of the second influencing factor, the second influencing factor refers to a factor affecting T2, and InputData2 represents a plurality of local temperature data of the target transformer.

[0079] In the embodiments of the present application, the first influencing factor and the second influencing factor refer to factors affecting the temperature of the transformer, such as load, ambient temperature, etc., mainly reflecting external conditions and control parameters. The local temperature includes oil temperature (insulating oil temperature) and shell temperature.

[0080] Step 202, calculating the temperature rise rate Δv of the target transformer in Δt time T :

[0081]

[0082] wherein T'2 represents the overall temperature of the target transformer predicted last time, Δt = t2-t1, t2 represents the collection time point of the current InputData2, and t1 represents the collection time point of the previous InputData2.

[0083] Step 203, if Δv T is greater than the preset temperature rise rate threshold, it indicates that the temperature rise of the target transformer is abnormal.

[0084] In the embodiments of the present application, the temperature of the target transformer (such as oil temperature, shell temperature) is an important indicator reflecting the running state and health condition of the target transformer, and is also a pre-stage reflection of problems such as insulation aging and failure of the heat dissipation system of the target transformer. Considering the influence of various influencing factors on the temperature of the target transformer, the temperature change of the target transformer can be more accurately predicted. Through the above steps 201 to 203, the monitoring and early warning of abnormal temperature rise are realized, and the accuracy of abnormal diagnosis and early warning of the target transformer is improved.

[0085] According to the preprocessed current local temperature data, the state of the target transformer is evaluated, and after the above step 201, the following steps 301 to 302 are further included. Wherein:

[0086] Step 301, according to multiple influencing factors θ i and the current InputData2, based on the Bayesian model, P(T2>T threshold , θ i |InputData2) is determined according to the following formula:

[0087]

[0088] wherein P(InputData2|T2>T threshold , θ i ) represents the likelihood of observing InputData under T2 greater than T threshold and multiple second influencing factors θ i , and P(T2>T threshold , θ i ) represents the likelihood of T2 being greater than Tthreshold T2 and the plurality of second influencing factors θ i the prior probability, P (InputData2) represents the marginal likelihood of InputData2, P (T2>T threshold , θ i |InputData2) represents the posterior probability of T2 greater than T threshold and the plurality of second influencing factors θ i ; wherein T threshold is a set temperature threshold.

[0089] Step 302, if P (T2>T threshold , θ i |InputData2) is greater than a set probability threshold, it indicates that the temperature of the target transformer is abnormal.

[0090] In the embodiments of the present application, through steps 301 to 302, the dynamic risk assessment of the overall temperature T2 of the target transformer is realized.

[0091] The current state data is preprocessed, specifically including:

[0092] The single discharge pulse of the current partial discharge signal is analyzed in time domain, frequency domain and time-frequency domain, and the time domain features, frequency domain features and time-frequency features of the partial discharge signal are extracted.

[0093] In the embodiments of the present application, the time domain features of the partial discharge signal include but are not limited to the amplitude, frequency, phase and energy density of the partial discharge signal, the frequency domain features include but are not limited to the center frequency, harmonic component, power spectral density, spectrum shape, bandwidth, etc. The time-frequency features include but are not limited to time-frequency distribution graph, instantaneous frequency and instantaneous amplitude, marginal spectrum, average power spectral density, time-frequency entropy, etc.

[0094] According to the preprocessed current partial discharge signal, the state of the target transformer is evaluated, including the following steps 401 to 402. Wherein:

[0095] Step 401, input the extracted time domain features, frequency domain features and time-frequency features of the partial discharge signal into the constructed insulation fault assessment model, and the insulation threat level corresponding to the insulation fault assessment model is output.

[0096] In the embodiments of the present application, the time domain features, frequency domain features and time-frequency features of the partial discharge signal sample can be directly associated with the insulation threat level, the critical values (thresholds) of different insulation threat levels are set in combination with standards (such as IEEE C57.104 or IEC 60270), each partial discharge signal sample is divided into different insulation threat levels, and then the partial discharge signal sample with the insulation threat level is used to train the insulation fault assessment model.

[0097] In step 402, if the insulation threat level output by the insulation fault assessment model is greater than the preset insulation threat level threshold, it indicates that the insulation performance of the target transformer is abnormal.

[0098] In the embodiments of the present application, through the above steps 401 to 402, the potential threat of the insulation performance of the target transformer can be evaluated and warned, and the accuracy of the abnormal diagnosis and warning of the target transformer is improved.

[0099] According to the preprocessed current local vibration signal, the state of the target transformer is evaluated, including the following steps 501 to 502. Among them:

[0100] In step 501, if at least part of the local vibration signals in the current multiple local vibration signals have a frequency that is an integer multiple of 50Hz and an amplitude greater than a preset first amplitude threshold, it indicates that the winding of the target transformer is loose.

[0101] In the embodiments of the present application, the above-mentioned integer multiple refers to a multiple of 50Hz and the multiple is an integer. For example, 1 times, 2 times or 3 times, etc., corresponding to at least part of the local vibration signals with a frequency of 50Hz, 100Hz or 150Hz, etc.

[0102] In step 502, if at least part of the local vibration signals in the current multiple local vibration signals have a frequency greater than 20Hz and less than 50Hz and an amplitude greater than a preset second amplitude threshold, it indicates that the core of the target transformer is loose.

[0103] In the embodiments of the present application, multiple vibration sensors are arranged at the top of the winding, near the core and on the surface of the box of the target transformer to collect vibration signals, i.e. local vibration signals, at different positions. Winding looseness fault usually shows more significantly at vibration sensors close to the winding, while core looseness fault shows more significantly at vibration sensors close to the core. Through vibration spectrum analysis, it is found that the vibration signal of winding looseness shows abnormally high amplitude near 50Hz or its integer multiple frequency, while core looseness mainly shows enhanced vibration peak in the low frequency band (usually 20Hz-50Hz). Therefore, through experiments, the first amplitude threshold for judging the winding abnormality of the transformer and the second amplitude threshold for judging the core abnormality of the transformer can be determined. Through the above steps, the potential threat of winding looseness and core looseness can be realized, and the accuracy of the abnormal diagnosis and warning of the target transformer is improved.

[0104] According to the preprocessed current grounding current, the state of the target transformer is evaluated, including:

[0105] According to the current grounding current, a first fault identification model based on attention convolutional neural network or reinforcement learning is used to identify the grounding fault and insulation damage fault of the target transformer.

[0106] In the embodiments of the present application, the convolutional neural network can extract local features and global features of complex spatio-temporal grounding current data through convolutional layers and pooling layers. The introduction of the attention mechanism in the convolutional neural network can make the convolutional neural network pay more attention to key features in the grounding current data, thereby improving the accuracy of fault prediction. The main role of reinforcement learning is dynamic adaptation, which continuously adjusts the prediction strategy through learning to adapt to the complexity and uncertainty of power system fault modes. Through continuous data learning and optimization, it can more sensitively identify minor abnormalities in the grounding current, such as weak harmonics and non-periodic fluctuations, thereby improving the accuracy of grounding current monitoring. Thus, the accuracy of the entire comprehensive monitoring and early warning method is improved to some extent.

[0107] According to the preprocessed current oil chromatographic data, the state of the target transformer is evaluated, including:

[0108] According to the current oil chromatographic data, a second machine learning model is used to identify overheat faults or discharge faults of the target transformer and the fault positions thereof, wherein the second machine learning model includes a GA-SVM model or a CEABC-WNN model, the GA-SVM model refers to a support vector machine SVM optimized by a genetic algorithm GA and trained, and the CEABC-WNN model refers to a wavelet neural network WNN model optimized by a chaotic enhanced artificial bee colony algorithm CEABC and trained.

[0109] In the embodiments of the present application, the oil chromatographic real-time monitoring is combined with the GA-SVM model, the parameter combination of the support vector machine SVM is optimized by the genetic algorithm GA, the SVM classifies and identifies overheat faults and discharge faults according to the components and concentrations of the dissolved gases in the oil of the target transformer monitored by the oil chromatographic real-time monitoring, so as to improve the accuracy and generalization ability of the early warning. The CEABC improves the global search ability and convergence speed of the algorithm by introducing the chaos theory to optimize the search process. The WNN model automatically extracts fault features by using the time-frequency localization characteristics of the wavelet transform. Firstly, the initial weights and thresholds of the WNN model are optimized by the CEABC to reduce the influence of artificial parameter setting, which can speed up the training process and help the WNN model to better converge to the global optimal solution or a better local optimal solution. Subsequently, the WNN model is trained by using the initial weights and thresholds optimized by the CEABC, so that it can accurately map the input features to the output fault type (thermal / electric). During the training process, the parameters of the WNN model are continuously adjusted by comparing the fault type output by the WNN model with the actual fault label of the input features, until the classification accuracy of the thermal fault and the electric fault is satisfactory, and the final WNN model is trained.

[0110] Optionally, in other embodiments of the present application, the pre-processing of the current state data further comprises:

[0111] The current partial discharge signal is sequentially subjected to first amplification, mixing, first filtering, second amplification, and second filtering. The single discharge pulse of the partial discharge signal after the second filtering is subjected to time domain analysis, frequency domain analysis, and time-frequency analysis, to extract the time domain feature, frequency domain feature, and time-frequency feature of the partial discharge signal.

[0112] In the embodiments of the present application, the role of mixing is to convert the partial discharge signal to intermediate frequency (IF) or lower frequency for subsequent processing. The signal after mixing usually contains multiple frequency components, and the first filtering can remove the noise and interference signals introduced in the mixing process, thereby improving the signal-to-noise ratio. The filter used for the first filtering is not specifically limited and can be selected as needed. For example, a program-controlled filter (such as wavelet denoising, adaptive filtering, narrowband interference elimination, etc.) can be used, and the parameters of the program-controlled filter can be dynamically adjusted as needed, so as to selectively extract the useful frequency band in the partial discharge signal after mixing. The purpose of the second amplification after the first filtering is to further improve the signal-to-noise ratio of the partial discharge signal, so that the partial discharge signal is clearer and more reliable. The filter used for the second filtering is not specifically limited and can be selected as needed. For example, a band-pass filter is used to ensure that the useful frequency band in the partial discharge signal after the second amplification is extracted.

[0113] The current grounding current is sequentially subjected to third filtering, smoothing, adjusting data offset, and normalization.

[0114] In the grounding current monitoring, due to the differences in characteristics (such as sensitivity, measurement range) of the monitoring device and the changes in environmental conditions (such as temperature, humidity, electromagnetic interference), the collected data often has problems of offset and different scales. In order to adjust the data offset and unify the data scale, data calibration is required, including offset correction and scale normalization. The purpose of offset correction is to eliminate the measurement reference point offset caused by the inherent deviation of the device or environmental factors. The average value or median of the monitoring data of a plurality of monitoring points set by calculation can be used to adjust the original data, for example, the original data is subtracted by the average value or median, so as to eliminate the system error. The Z-score method is used for scale normalization, which converts the data into a standard normal distribution form, which is suitable for cases where the data range may change.

[0115] The current partial temperature data is sequentially subjected to wavelet denoising and error correction to eliminate the influence of environmental interference and the characteristics of the temperature monitoring device itself on the measurement results.

[0116] The current oil chromatogram data (the concentration data of each gas component) is cleaned by using median replacement and K-neighbor interpolation method to eliminate noise and outliers.

[0117] The current local vibration signal is sequentially subjected to fourth filtering and blind source separation processing to improve the signal-to-noise ratio of the local vibration signal.

[0118] Optionally, in other embodiments of the present application, the collected partial discharge signal is subjected to mixing processing, including sequentially performing frequency reduction (local oscillator signal), sampling, comparison (comparison with reference signal), and interference elimination.

[0119] In the embodiments of the present application, frequency reduction refers to multiplying the partial discharge signal RF and the local oscillator signal (local oscillator signal) LO to generate a signal containing sum frequency (RF+LO) and difference frequency (|RF-LO|). The local oscillator signal LO is not specifically limited and can be selected as needed.

[0120] Optionally, in other embodiments of the present application, the current grounding current is subjected to smoothing processing by using a sliding average filtering method to reduce data fluctuation and improve data stability.

[0121] Optionally, in other embodiments of the present application, the third filtering includes at least one of low-pass digital filtering and Kalman filtering to remove high-frequency noise and random interference.

[0122] Optionally, in other embodiments of the present application, the fourth filtering adopts an adaptive filtering algorithm.

[0123] Optionally, according to the preprocessed current partial discharge signal, the state of the target transformer is evaluated, and the method further includes:

[0124] The preprocessed current multiple partial discharge signals collected by the constructed multi-dimensional sensor network are subjected to multi-discharge fault point positioning and displayed in a three-dimensional space.

[0125] The discharge position, displacement, intensity change, and discharge type change of the multi-discharge fault point are tracked and recorded in real time by using a dynamic tracking algorithm to ensure comprehensive control of the discharge phenomenon.

[0126] In the embodiments of the present application, the multi-discharge fault point positioning is achieved by using a multi-dimensional sensor network constructed by combining various sensors such as electroacoustic, electro-electric, and acoustic-acoustic sensors, and by using time difference positioning and beam forming technologies to realize three-dimensional positioning of the multi-discharge fault point. The electroacoustic sensor combination refers to the integrated use of a sound sensor (microphone) and an electric sensor (such as a current or voltage sensor). The electro-electric sensor combination refers to the joint use of different types of electrical sensors (such as voltage, current, and power sensors). The acoustic-acoustic sensor combination refers to the joint use of multiple sound sensors (such as a microphone array).

[0127] Optionally, in other embodiments of the present application, the current state data is preprocessed, and the preprocessing further includes:

[0128] The local vibration signals processed by the blind source separation are analyzed in time domain, frequency domain and time-frequency domain, the frequency complexity FCA, vibration stationarity DET, vibration correlation MPC and energy similarity EDR of the local vibration signals are extracted, and the permutation entropy PE of the state data of the target transformer is calculated.

[0129] In the embodiments of the present application, the time domain analysis refers to analyzing the local vibration signals directly in the time domain, focusing on the time domain characteristics of the signals such as amplitude, root mean square (RMS), kurtosis, peak factor, etc., to quickly detect abnormal vibration. The frequency domain analysis refers to converting the local vibration signals from the time domain to the frequency domain by Fourier transform (FFT) to obtain a frequency spectrum. Through the frequency spectrum, specific frequency components of core loosening such as low-frequency vibration can be identified. The frequency domain analysis can reveal the frequency components and distribution of the vibration signals. The time-frequency analysis refers to analyzing the vibration signals in the time domain and frequency domain by using time-frequency analysis methods such as wavelet transform, short-time Fourier transform (STFT) and Hilbert-Huang transform (HHT) to obtain a time-frequency graph. The results of time domain analysis, frequency domain analysis and time-frequency analysis can be combined to make a comprehensive diagnosis of mechanical faults. For example, time domain analysis can quickly detect abnormal vibration, frequency domain analysis can identify specific vibration fault frequencies, and time-frequency analysis can capture the non-stationary characteristics of the vibration signals. Through multi-method fusion, the vibration fault location and type can be more accurately located. The frequency complexity reflects the richness of different frequency components in the signal, which is an important indicator for judging the internal mechanical state of the target transformer. Changes in vibration stationarity may indicate loosening or wear of internal mechanical components. By calculating the vibration correlation of vibration signals at different positions, i.e. different local vibration signals, the interaction and coordination between internal mechanical components of the target transformer can be judged. The energy similarity is used to evaluate the similarity of the energy distribution of the local vibration signals at different time periods. By comparing the energy distribution of the local vibration signals, the mechanical stability of the target transformer under different states can be judged.

[0130] Optionally, in other embodiments of the present application, according to the preprocessed current local vibration signals, the state of the target transformer is evaluated, and the evaluation further includes:

[0131] Before step 501, if at least one of the following conditions occurs: the FCA of the current local vibration signal is greater than a preset FCA threshold, the DET is less than a preset DET threshold, the MPC is less than a preset MPC threshold, the EDR is greater than a preset EDR threshold, and the PE is greater than a preset PE threshold, it indicates that the target transformer may have a mechanical fault, and the mechanical fault is further judged according to steps 501 to 502.

[0132] In the embodiments of the present application, the FCA fault threshold, the DET fault threshold, the MPC fault threshold, the EDR fault threshold and the PE fault threshold are set according to historical data and experience. For example, according to historical data and experience, it is concluded that the normal range of FCA is 0.5-1.0, the normal range of DET is 0.7-1.0, the normal range of MPC is 0.8-1.0, the normal range of EDR is 0.5-0.8, and the normal range of PE is 0.5-0.8, then the FCA fault threshold is set to be greater than 1.0, the DET fault threshold is set to be less than 0.7, the MPC fault threshold is set to be less than 0.8, the EDR fault threshold is set to be less than 0.5, and the PE fault threshold is set to be greater than 0.8. If the FCA exceeds the FCA threshold, it indicates that the frequency complexity of the local vibration signal is high, and there may be mechanical failure. If the DET is lower than the DET threshold, it indicates that the certainty and predictability of the local vibration signal is poor, and there may be mechanical failure. If the MPC is lower than the MPC threshold, it indicates that the mutual prediction ability between the local vibration signals is poor, and there may be mechanical failure. If the EDR exceeds the EDR threshold, it indicates that the energy distribution of the local vibration signal in different time periods is large, and there may be mechanical failure. If the PE exceeds the threshold, it indicates that the complexity and randomness of the state data of the target transformer is high, and there may be mechanical failure. By comprehensively analyzing the time-frequency characteristics of the local vibration signal and the permutation entropy value of the state data of the target transformer, the mechanical stability state of the target transformer can be effectively evaluated and failure discrimination can be performed. In combination with FCA, DET, MPC, EDR and PE, the state of the target transformer can be more comprehensively evaluated, potential mechanical failure can be found in time, and the reliable operation of the target transformer can be ensured.

[0133] Optionally, in other embodiments of the present application, the current state data is preprocessed, and the preprocessing specifically further includes:

[0134] After the current oil chromatogram data (the concentration data of each gas component) is cleaned by the median replacement and K-nearest neighbor interpolation method, the oil chromatogram sample data is subjected to DGA analysis, and the key failure features of the oil chromatogram sample data are extracted.

[0135] In the embodiments of the present application, the key failure features of the oil chromatogram sample data extracted by the DGA analysis method refer to the concentrations of the dissolved gases in the transformer oil and the concentration ratio values extracted by the DGA analysis method. The dissolved gases in the transformer oil include but are not limited to H2, CH4, CO, C2H6, CO2, C2H4, C2H2, etc., and the concentration ratio values of the dissolved gases in the transformer oil include but are not limited to C2H2 / C2H4 (acetylene to ethylene concentration ratio), CH4 / H2 (methane to hydrogen concentration ratio) and C2H4 / C2H6 (ethylene to ethane concentration ratio).

[0136] According to the current oil chromatographic data, a second machine learning model is used to identify the overheat fault or discharge fault of the target transformer and the fault location thereof, including:

[0137] The key fault features of the extracted oil chromatographic sample data are input into the second machine learning model, and the overheat fault or discharge fault of the target transformer and the fault location thereof are identified.

[0138] On the basis of using the second machine learning model to explicitly distinguish between thermal faults and electrical faults, the key fault features of the oil chromatographic sample data are extracted in combination with the corresponding DGA (oil chromatographic analysis), so that the thermal fault and electrical fault machine positions are more accurately distinguished.

[0139] Optionally, in other embodiments of the present application, the process of optimizing the initial parameters of the WNN model by using the CEABC includes the following steps (1) to (11). Wherein:

[0140] Step (1), define the structure of the WNN model, including the number of nodes of the input layer, the wavelet hidden layer and the output layer.

[0141] Step (2), initialize the parameters of the CEABC algorithm, including the number of honey bee populations and the number of iterations.

[0142] Step (3), initialize the weights and thresholds (initial values of each node) of the WNN model. The weights are the parameters connecting the hidden layer (wavelet basis function layer) and the output layer, which determine the influence degree of the hidden layer output on the final output.

[0143] Step (4), encode the weights and thresholds of the WNN model into food sources in the CEABC algorithm.

[0144] Step (5), for each food source, calculate the loss function of the corresponding WNN model on the training data, and take the loss function as the fitness.

[0145] Step (6), each employed bee searches for a new food source (i.e. new weights and thresholds) near its corresponding food source, and evaluates the fitness of the searched new food source.

[0146] Step (7), if the fitness of the new food source is better than that of the old food source, replace the old food source.

[0147] Step (8), according to the fitness proportion of the food sources, the follower bees are distributed, and the follower bees select the food sources with high fitness for local search.

[0148] Step (9), if a certain food source is not improved after multiple iterations, it is considered to be exhausted (falling into a local optimum), at which time the scout bee randomly selects a new food source to replace the exhausted food source, i.e., the scout bee re-explores a new food source.

[0149] Step (10), using a chaotic mapping to update the position of the food source to increase the randomness of the search.

[0150] Step (11), repeating the above steps (5) to (10) until a preset termination condition (such as the maximum number of iterations) is reached, outputting the optimal food source and its fitness value, and the initial weights and thresholds of the optimized WNN.

[0151] In the embodiments of the present application, the iteration of the CEABC (repeated calculation of steps (5) to (10)) and the optimization process combine multi-strategy fusion and adaptive adjustment, which adaptively adjusts according to the actual situation in the search process, making the algorithm more flexible and efficient.

[0152] Optionally, in other embodiments of the present application, according to the preprocessed current grounding current, the state of the target transformer is evaluated, further comprising:

[0153] According to the preprocessed historical grounding current data and the preprocessed current grounding current data of the target transformer, the future grounding current is predicted using a Bi-LSTM algorithm;

[0154] The future grounding current is compared with the currently collected grounding current to determine the deviation therebetween;

[0155] If the deviation between the future grounding current and the currently collected grounding current exceeds a set value, it indicates that the target transformer is abnormally grounded.

[0156] In the embodiments of the present application, through the above steps, the potential grounding fault can be evaluated and warned, and the monitoring and warning accuracy of the state of the target transformer is improved.

[0157] Optionally, in other embodiments of the present application, according to the preprocessed current state data, the state of the target transformer is evaluated from multiple angles, further comprising steps 601 to 605. Wherein:

[0158] Step 601, calculating the correlation coefficient between the preprocessed current state data.

[0159] In the embodiments of the present application, the correlation coefficient between the state data is calculated, i.e., the correlation coefficient between any two variables is calculated, and the type of the correlation coefficient is not specifically limited and can be selected as needed.

[0160] Step 602, constructing a correlation coefficient matrix according to correlation coefficients between state data.

[0161] Step 603, constructing an undirected graph network based on the correlation coefficient matrix, wherein a name of a node of the undirected graph network is a name of each state data (such as oil chromatographic data, partial discharge signal, grounding current, local temperature data and local vibration signal), an attribute value of the node is a corresponding state data value, and an edge is a connection line between a pair of significantly correlated state data.

[0162] In the embodiments of the present application, the correlation coefficient threshold is not specifically limited and can be set by the actual demand, for example, the correlation coefficient threshold is set to 0.5.

[0163] Step 604, identifying an independent community in the undirected graph network through a community detection algorithm.

[0164] Step 605, identifying an abnormal state of the target transformer according to the identified independent community.

[0165] In the embodiments of the present application, the abnormal state of the target transformer can be identified according to the identified independent community through community structure analysis. For example, the connectivity and stability inside each independent community are checked, and if an abnormal disconnection or a large number of newly added edges appear inside an independent community, it may indicate that the independent community has a local problem. The abnormal disconnection means that at least one node inside the independent community loses all edges connected thereto, and the large number of newly added edges means that the number of edges newly added to at least one node inside the independent community exceeds a set edge number threshold. The edge number threshold is not specifically limited in the embodiments of the present application and can be set by the actual demand. The abnormal state of the target transformer can be identified according to the identified independent community through analysis of changes in the attribute values of the nodes. For example, the changes in the attribute values of the nodes of each independent community over time are observed, and if the attribute values of some nodes exceed a preset safe operation threshold (for example, a certain local temperature data value exceeds the safe operation threshold), it means that these nodes have failed. The attribute values of the nodes of each independent community can also be compared with the attribute values of the nodes of the independent community during the historical normal operation, and if the difference obtained by the comparison is greater than a preset difference threshold, it means that the difference deviates from the normal state significantly, and the node whose difference is greater than the preset difference threshold has a corresponding failure.

[0166] Optionally, in other embodiments of the present application, the multi-angle evaluation of the state of the target transformer according to the preprocessed current state data further includes:

[0167] adopting a hierarchical clustering model based on a bottom-up aggregation strategy to cluster the preprocessed current state data;

[0168] According to the clustering result, comprehensive fault diagnosis is performed, and the fault of the target transformer is identified, the fault including at least one of a partial discharge fault, a mechanical fault, a grounding fault, an insulation fault, and an overheat fault.

[0169] In the embodiments of the present application, the state data includes oil chromatographic data, partial discharge signals, grounding currents, local temperature data, and local vibration signals. First, the sample data set is cleaned and standardized, and then each data point (each monitoring data) is taken as an initial cluster. The similarity between clusters is calculated by using the Euclidean distance, and the most similar clusters are gradually merged to form a higher level of cluster structure. This process is iterated until the number of clusters reaches the optimal value or the similarity between clusters is lower than a threshold. In the clustering process, the Calinski-Harabasz index evaluation index is combined to monitor the clustering effect, and the parameters of the hierarchical clustering model are continuously adjusted to optimize the clustering result. Taking each data point as an initial cluster avoids the subjectivity of manually presetting the number of clusters or the cluster center, provides rich initial information for subsequent hierarchical clustering, and helps to discover the potential structure and pattern in the data. The hierarchical clustering model can merge data points with similar fault characteristics into the same cluster by gradually iterating merging, Euclidean distance calculation, and Calinski-Harabasz evaluation, thereby realizing rapid identification and positioning of fault categories.

[0170] Optionally, in other embodiments of the present application, the transformer comprehensive monitoring and early warning method described above further includes the following steps 701 to 704. Among them:

[0171] Step 701, a digital twin model of the target transformer is constructed.

[0172] In the embodiments of the present application, the multi-physics field simulation (multi-source simulation data of the transformer under different working conditions obtained by finite element analysis (FEA) and the like (oil chromatographic data, partial discharge signals, local temperature data, local vibration signals, grounding currents)), the measured multi-source data of the transformer, and the historical operation data of the transformer (including historical operation data such as operation records, maintenance records, and fault records) are taken as the original data pool to establish the digital twin model of the target transformer.

[0173] Step 702, the running process of the target transformer is simulated synchronously through the digital twin model, and the state data of the digital twin model is collected in real time.

[0174] Step 703, an input sequence is input into the time series difference learning model, and the future state data of the target transformer is predicted through the time series difference learning model; wherein the input sequence includes: a plurality of pairs of input data, each pair of input data including the state data of the digital twin model and the target transformer at the same time.

[0175] For example, assuming that there are n pairs of input data in the input sequence, let the current time be T m , the input sequence exemplarily can include:

[0176] Data(T z ), Data'(T z ), Data(T z +1), Data'(T z +1), …, Data(T m ), Data'(T m ), wherein Data(T i ) represents the state data of the target transformer at T i , Data'(T i ) represents the state data of the digital twin model at T i , i is an integer not less than z and not greater than m, and z = m-n+1.

[0177] In step 704, the future state data of the target transformer is taken as the current state data of the target transformer, and the current state data is preprocessed and subsequent steps are performed to prewarn each abnormal state of the target transformer in the future.

[0178] In the embodiment of the present application, during the training process of the time series differential learning model, the mean square error MSE is used as the loss function to evaluate the difference between the predicted value and the actual value, and the model parameters are optimized through the back propagation algorithm. When the digital model is dynamically updated, the time series differential learning model is retrained or fine-tuned according to the newly obtained measured data or simulation results, so as to continuously train and improve the time series differential learning model, so that its generalization ability becomes stronger and stronger, thereby improving the prediction accuracy, and the retrained or fine-tuned time series differential learning model is used to predict the dynamics of the target transformer.

[0179] Optionally, in other embodiments of the present application, the transformer comprehensive monitoring and prewarning method further comprises:

[0180] According to the evaluation results obtained by the multi-angle evaluation of the state of the target transformer, a state evaluation report is generated.

[0181] In the embodiment of the present application, the form of the generated state evaluation report can be selected by itself, including but not limited to forms such as table, chart, text description, interactive report, etc., and the above forms can also be combined to ensure the comprehensiveness and readability of the information. The state evaluation report contains the real-time collected state data of the target transformer, the multi-angle evaluation results of the target transformer, the prewarning measures and the maintenance suggestions.

[0182] The formulation of the maintenance suggestion includes:

[0183] 1) Priority, prioritize multiple faults (abnormalities), ensure that critical faults are given priority;

[0184] 2) Provide specific maintenance solutions for different types of faults, according to relevant standards or maintenance manuals;

[0185] 3) Risk assessment, assess the risks that may be encountered during maintenance, and provide corresponding preventive measures, according to relevant standards or maintenance manuals.

[0186] Based on the same inventive concept, the embodiments of the present application also provide a transformer comprehensive monitoring and early warning system for implementing the transformer comprehensive monitoring and early warning method described above. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme described in the above method, so the specific limitations in one or more transformer comprehensive monitoring and early warning system embodiments provided below can be referred to the limitations of the transformer comprehensive monitoring and early warning method in the above, which will not be repeated here.

[0187] In an exemplary embodiment, as shown in Figure 2 a transformer comprehensive monitoring and early warning system 80 is provided, comprising:

[0188] The acquisition module 801 is configured to acquire the state data of the target transformer in real time, and the state data includes oil chromatographic data, partial discharge signals, ground current, local temperature data and local vibration signals.

[0189] The data processing module 802 is configured to preprocess the current state data.

[0190] The multi-angle evaluation module 803 is configured to evaluate the state of the target transformer from multiple angles according to the preprocessed current state data.

[0191] The early warning module 804 is configured to early warn each abnormal state of the target transformer if the state of the target transformer is abnormal according to at least one angle evaluation.

[0192] In the embodiments of the present application, the data processing module 802 can be a hardware chip with data processing function, the multi-angle evaluation module 803 is a server or terminal host connected with the data processing module 802 through a switch, and the switch can also include a backup switch. The transformer comprehensive monitoring and early warning system further comprises a data storage system connected with the switch and located on the same layer as the terminal host, and the stored data is as follows:

[0193] The partial discharge data, ground current data, partial vibration data, oil chromatogram data, and partial temperature data collected in the past two years are stored. When the partial discharge data is normal, the storage interval is less than or equal to 1 minute; when the ground current data is normal, the data storage interval is less than or equal to 1 minute, and when the ground current data is abnormal, all data should be recorded; when the partial vibration data is normal, the storage interval is less than or equal to 1 minute, and when the partial vibration data is abnormal, all data should be recorded; the partial temperature data detection interval is less than or equal to 1 hour, the oil chromatogram data storage interval is less than or equal to 1 day, and all stored data contains time information accurate to seconds.

[0194] The process of multi-angle evaluation of the state of the target transformer is described in the above embodiment of the transformer comprehensive monitoring and early warning method, which will not be repeated here.

[0195] Optionally, in other embodiments of the present application, the data processing module 802 is further configured to:

[0196] The current partial discharge signal is sequentially subjected to first amplification, mixing, first filtering, second amplification, and second filtering. The single discharge pulse of the partial discharge signal after the second filtering is subjected to time domain analysis, frequency domain analysis, and time-frequency analysis, and the time domain feature, frequency domain feature, and time-frequency feature of the partial discharge signal are extracted;

[0197] The current ground current is sequentially subjected to third filtering, smoothing, adjustment of data offset, and normalization;

[0198] The current partial temperature data is sequentially subjected to wavelet denoising and error correction to eliminate the influence of environmental interference and the characteristics of the temperature monitoring device itself on the measurement results;

[0199] The current oil chromatogram data (concentration data of each gas component) is cleaned by using the median replacement and K-nearest neighbor interpolation method to eliminate noise and outliers;

[0200] The current partial vibration signal is sequentially subjected to fourth filtering and blind source separation processing to improve the signal-to-noise ratio of the partial vibration signal.

[0201] Optionally, in other embodiments of the present application, the multi-angle evaluation module 803 is further configured to:

[0202] The current multiple partial discharge signals collected by the constructed multi-dimensional sensor network after preprocessing are subjected to multi-discharge fault point positioning and displayed in a three-dimensional space;

[0203] The discharge position, displacement, intensity change, and discharge type change of the multi-discharge fault point are tracked and recorded in real time by using a dynamic tracking algorithm, ensuring comprehensive control of the discharge phenomenon.

[0204] Optionally, in other embodiments of the present application, the data processing module 802 is further configured to:

[0205] The local vibration signal processed by the blind source separation is subjected to time domain analysis, frequency domain analysis and time-frequency analysis, and the frequency complexity FCA, vibration stationarity DET, vibration correlation MPC and energy similarity EDR of the local vibration signal are extracted, and the permutation entropy PE of the state data of the target transformer is calculated.

[0206] Correspondingly, the multi-angle evaluation module 803 is further configured to:

[0207] Before step 501, if at least one of the following conditions occurs, i.e., the FCA of the current local vibration signal is greater than a preset FCA threshold, the DET is less than a preset DET threshold, the MPC is less than a preset MPC threshold, the EDR is greater than a preset EDR threshold, and the PE is greater than a preset PE threshold, it indicates that the target transformer may have a mechanical fault, and the mechanical fault is further determined according to steps 501 to 502.

[0208] Optionally, in other embodiments of the present application, the data processing module 802 is further configured to:

[0209] After the current oil chromatogram data (the concentration data of each gas component) is cleaned by the median replacement and K-nearest neighbor interpolation method, the oil chromatogram sample data is subjected to DGA analysis, and the key fault features of the oil chromatogram sample data are extracted.

[0210] Correspondingly, the multi-angle evaluation module 803 is further configured to:

[0211] The key fault features of the extracted oil chromatogram sample data are input into the second machine learning model, and the overheat fault or discharge fault of the target transformer and the fault position thereof are identified.

[0212] Optionally, in other embodiments of the present application, the multi-angle evaluation module 803 is further configured to:

[0213] The initial parameters of the WNN model are optimized by the CEABC according to the following process:

[0214] The structure of the WNN model is defined, including the number of nodes of the input layer, the wavelet hidden layer and the output layer;

[0215] The parameters of the CEABC algorithm are initialized, including the number of honey bee populations and the number of iterations;

[0216] The weights and thresholds (initial values of each node) of the WNN model are initialized, and the weights are parameters connecting the hidden layer (wavelet basis function layer) and the output layer, which determine the influence degree of the hidden layer output on the final output;

[0217] The weights and thresholds of the WNN model are encoded into food sources in the CEABC algorithm;

[0218] For each food source, calculate the loss function of its corresponding WNN model on the training data, and take the loss function as the fitness;

[0219] Each employed bee searches for a new food source (i.e., new weights and thresholds) near its corresponding food source, and evaluates the fitness of the searched new food source;

[0220] If the fitness of the new food source is better than that of the old food source, replace the old food source;

[0221] According to the fitness ratio of the food sources, assign the follower bees to select the food source with high fitness for local search;

[0222] If a certain food source has not been improved after multiple iterations, it is considered to have been exhausted (falling into a local optimum), at which time the scout bee is called to randomly select a new food source to replace the exhausted food source, i.e., the scout bee is called to re-explore a new food source;

[0223] Use chaotic mapping to update the position of the food source to increase the randomness of the search;

[0224] Repeat the above steps of calculating the loss function of the corresponding WNN model on the training data for each food source, and taking the loss function as the fitness, to the step of using chaotic mapping to update the position of the food source, until a preset termination condition (such as the maximum number of iterations) is reached, outputting the optimal food source and its fitness value, and the initial weights and thresholds of the optimal WNN.

[0225] Optionally, in other embodiments of the present application, the multi-angle evaluation module 803 is further used for:

[0226] According to the historical grounding current data and the currently collected grounding current data of the target transformer, the future grounding current is predicted by using a Bi-LSTM algorithm;

[0227] The future grounding current is compared with the currently collected grounding current to determine the deviation therebetween;

[0228] If the deviation between the future grounding current and the currently collected grounding current exceeds a set value, it indicates that the target transformer is grounded abnormally.

[0229] Optionally, in other embodiments of the present application, the multi-angle evaluation module 803 is further used for:

[0230] The correlation coefficients between the preprocessed current state data are calculated;

[0231] According to the correlation coefficients between the state data, a correlation coefficient matrix is constructed;

[0232] Based on the correlation coefficient matrix, an undirected graph network is constructed, wherein the name of a node of the undirected graph network is the name of each state data (such as oil chromatographic data, partial discharge signal, grounding current, local temperature data and local vibration signal), the attribute value of the node is the corresponding state data value, and the edge is a connection line between a pair of significantly correlated state data.

[0233] An independent community in the undirected graph network is identified through a community detection algorithm.

[0234] According to the identified independent community, an abnormal state of the target transformer is identified.

[0235] Optionally, in other embodiments of the present application, the multi-angle evaluation module 803 is further used to:

[0236] A hierarchical clustering model based on a bottom-up aggregation strategy is used to cluster the preprocessed current state data.

[0237] According to the clustering result, a comprehensive fault diagnosis is performed to identify the fault of the target transformer, which includes at least one of a partial discharge fault, a mechanical fault, a grounding fault, an insulation fault and an overheating fault.

[0238] Optionally, in other embodiments of the present application, the multi-angle evaluation module 803 is further used to:

[0239] A digital twin model of the target transformer is constructed;

[0240] The operation process of the target transformer is simulated synchronously through the digital twin model, and the state data of the digital twin model is collected in real time;

[0241] The input sequence is input into the time series difference learning model, and the future state data of the target transformer is predicted through the time series difference learning model; wherein the input sequence includes: a plurality of pairs of input data, each pair of input data including the state data of the digital twin model and the target transformer at the same time, and the related description is detailed in the description of the above method embodiments;

[0242] The future state data of the target transformer is taken as the current state data of the target transformer, and the subsequent steps of pre-processing the current state data are performed to prewarn each abnormal state of the target transformer in the future.

[0243] Optionally, in other embodiments of the present application, the transformer comprehensive monitoring and early warning system described above further includes:

[0244] The report generation module 805 is configured to generate a state evaluation report according to the evaluation result obtained by the multi-angle evaluation of the state of the target transformer.

[0245] Optionally, in other embodiments of the present application, the collection module 801 comprises:

[0246] The temperature collection module 8011 is configured to collect the temperature of the target transformer in real time, including the oil temperature and the shell temperature.

[0247] The partial discharge collection module 8012 is configured to collect the partial discharge signal of the target transformer in real time.

[0248] The grounding current collection module 8013 is configured to collect the grounding current of the target transformer in real time.

[0249] The partial vibration collection module 8014 is configured to collect the partial vibration signal of the target transformer in real time.

[0250] The oil chromatogram collection module 8015 is configured to collect the oil chromatogram data of the target transformer in real time, and the oil chromatogram data refers to the gas component and the concentration data of each gas in the oil of the target transformer.

[0251] In an exemplary embodiment, a computer device, which can be a server or a terminal, has an internal structure as shown in Figure 3 The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store processing data of transformer comprehensive monitoring and early warning. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a transformer comprehensive monitoring and early warning method.

[0252] Those skilled in the art can understand that Figure 3 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0253] In an example embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0254] In an example embodiment, a computer readable storage medium is provided, storing a computer program, the computer program implementing the steps in the above method embodiments when executed by a processor.

[0255] In an example embodiment, a computer program product is provided, including a computer program, the computer program implementing the steps in the above method embodiments when executed by a processor.

[0256] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0257] It can be understood by those skilled in the art that all or part of the processes in the above method embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Any reference to memory, database or other medium used in the embodiments provided by the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0258] The database involved in each of the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, and the like, without being limited thereto. The processor involved in each of the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, and the like, without being limited thereto.

[0259] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but it should be considered that any combination of the technical features is within the scope of the present disclosure, as long as there is no contradiction.

[0260] The principles and implementation modes of the present application are described by applying specific examples herein, and the above embodiments are only used to help understand the method of the present application and its core idea. Meanwhile, for those skilled in the art, the specific implementation modes and application ranges will be changed according to the idea of the present application. In summary, the content of the present description should not be understood as a limitation of the present application.

Claims

1. A comprehensive monitoring and early warning method for transformers, characterized in that, The comprehensive transformer monitoring and early warning method includes: The status data of the target transformer is collected in real time, including oil chromatography data, partial discharge signal, grounding current, local temperature data, and local vibration signal; wherein, the local temperature data includes local oil temperature and local casing temperature. Preprocess the current state data; Based on the preprocessed current state data, the state of the target transformer is evaluated from multiple perspectives. If the target transformer is found to be in an abnormal state from at least one perspective, an early warning shall be issued for each abnormal state of the target transformer. The step of evaluating the state of the target transformer from multiple perspectives based on the preprocessed current state data includes: Based on the preprocessed local temperature data, assess whether the temperature rise and overall temperature of the target transformer are abnormal. Based on the preprocessed partial discharge signal, assess whether the insulation performance of the target transformer is abnormal; Based on the pre-processed local vibration signals, assess the winding loosening and core loosening of the target transformer; Based on the pre-processed grounding current, assess the grounding faults and insulation damage faults of the target transformer; Based on the pre-processed oil chromatography data, assess the overheating or discharge faults and fault locations of the target transformer. The step of assessing whether the temperature rise of the target transformer is abnormal based on the preprocessed local temperature data includes: Based on multiple secondary influencing factors and current Using parameter combinations The first machine learning model predicts the current overall temperature of the target transformer. ;in, }, This indicates the quantity of the second influencing factor, where the second influencing factor refers to the factor that affects... Factors, This represents multiple local temperature data points of the target transformer; Calculate the target transformer in heating rate over time : ； in, This represents the overall temperature of the target transformer as predicted in the previous analysis. , Indicates the current The time point of collection, Indicates the previous The time point of data collection; like If the temperature rise rate exceeds the preset threshold, it indicates that the temperature rise of the target transformer is abnormal. The step of assessing whether the overall temperature of the target transformer is abnormal based on the preprocessed local temperature data includes: Based on multiple secondary influencing factors and current Based on the Bayesian model, it is determined according to the following formula. : = ； in, Indicates greater than of and several secondary influencing factors The following observations Likelihood, Indicates greater than of and several secondary influencing factors The prior probability, express Marginal likelihood, Indicates greater than of and several secondary influencing factors The posterior probability; where, The set temperature threshold; like If the probability exceeds the set probability threshold, it indicates that the temperature of the target transformer is abnormal; The preprocessing of the current state data includes: The current partial discharge signal is sequentially amplified, mixed, filtered, amplified, and filtered. Then, time-domain, frequency-domain, and time-frequency analysis are performed on a single discharge pulse of the partial discharge signal after the second filtering process to extract the time-domain, frequency-domain, and time-frequency features of the partial discharge signal. The current grounding current is sequentially subjected to third filtering, smoothing, data offset adjustment, and normalization. The step of evaluating whether the insulation performance of the target transformer is abnormal based on the preprocessed partial discharge signal includes: The time-domain features, frequency-domain features, and time-frequency features of the extracted partial discharge signal are input into the constructed insulation fault assessment model, and the insulation fault assessment model outputs the corresponding insulation threat level. If the insulation threat level output by the insulation fault assessment model is greater than the preset insulation threat level threshold, it indicates that the insulation performance of the target transformer is abnormal. The assessment of winding loosening and core loosening of the target transformer based on the preprocessed local vibration signal includes: If at least some of the current local vibration signals have frequencies that are integer multiples of 50Hz and amplitudes that are greater than a preset first amplitude threshold, it indicates that the windings of the target transformer are loose. If, among the current plurality of local vibration signals, at least some of the local vibration signals have a frequency greater than 20Hz and less than 50Hz and an amplitude greater than a preset second amplitude threshold, it indicates that the core of the target transformer is loose. The step of identifying the abnormal state of the target transformer based on the preprocessed current state data includes: Calculate the correlation coefficients between the current state data after preprocessing; Construct a correlation coefficient matrix based on the correlation coefficients between the data of each state; Based on the correlation coefficient matrix, an undirected graph network is constructed, wherein the name of each node in the undirected graph network is the name of each state data, the attribute value of each node is the corresponding state data value, and the edge is the connecting line between significantly related state data pairs; wherein, if the correlation coefficient of two state data is greater than a preset correlation coefficient threshold, they are defined as a group of significantly related state data pairs. Independent communities in the undirected graph network are identified using a community detection algorithm; Based on the identified independent communities, the abnormal state of the target transformer is identified.

2. The transformer integrated monitoring and early warning method according to claim 1, characterized in that, Before performing a multi-faceted evaluation of the target transformer's state based on the preprocessed current state data, the method further includes: Build the first machine learning model and utilize the particle swarm optimization algorithm and as well as The parameters of the first machine learning model are optimized, wherein, This provides multiple local temperature data for the sample transformer. The overall temperature of the sample transformer. As input to the first machine learning model, As the output of the first machine learning model, the optimal parameter combination is found through iterative search: ； in, This represents the first machine learning model. This indicates the primary influencing factor, among which, }, This indicates the quantity of the first influencing factor, where the first influencing factor refers to the factor that affects... Factors, This represents the prediction error function of the first machine learning model. This represents the prediction error of the first machine learning model. The minimum parameter combination of the first machine learning model. This represents the optimal parameter combination found by the particle swarm optimization algorithm.

3. The transformer integrated monitoring and early warning method according to claim 1, characterized in that, The assessment of grounding faults and insulation damage faults of the target transformer based on the pre-processed grounding current includes: Based on the preprocessed grounding current, the grounding fault and insulation damage fault of the target transformer are identified using a first fault identification model based on attention convolutional neural network or reinforcement learning.

4. The transformer integrated monitoring and early warning method according to claim 1, characterized in that, The step of evaluating the overheating or discharge fault and its location in the target transformer based on the pre-processed oil chromatography data includes: Based on the preprocessed oil chromatography data, a second machine learning model is used to identify the overheating fault or discharge fault and the fault location of the target transformer. The second machine learning model includes a GA-SVM model or a CEABC-WNN model. The GA-SVM model refers to the final SVM obtained by optimizing the parameter combination of the Support Vector Machine (SVM) using the Genetic Algorithm (GA) and training it. The CEABC-WNN model refers to the final WNN model obtained by optimizing the initial parameter combination of the Wavelet Neural Network (WNN) model using the Chaotic Enhanced Artificial Bee Colony Algorithm (CEABC) and training it.

5. The transformer integrated monitoring and early warning method according to claim 1, characterized in that, The step of evaluating the state of the target transformer from multiple perspectives based on the preprocessed current state data further includes: A hierarchical clustering model based on a bottom-up aggregation strategy is used to cluster the preprocessed current state data; Based on the clustering results, a comprehensive fault diagnosis is performed to identify the faults of the target transformer, including partial discharge faults, mechanical faults, grounding faults, insulation faults, and overheating faults.

6. The transformer integrated monitoring and early warning method according to any one of claims 1 to 5, characterized in that, Also includes: Construct a digital twin model of the target transformer; The operation process of the target transformer is simulated synchronously using the digital twin model, and the status data of the digital twin model is collected in real time. The input sequence is fed into a time-series differential learning model, which predicts the future state data of the target transformer. The input sequence includes multiple pairs of input data, each pair of input data including the state data of the digital twin model and the target transformer at the same time. The future state data of the target transformer is used as the current state data of the target transformer. Preprocessing of the current state data and subsequent steps are performed to provide early warning for each abnormal state of the target transformer in the future.

7. A comprehensive transformer monitoring and early warning system, characterized in that, The transformer integrated monitoring and early warning system includes: The acquisition module is used to acquire the current status data of the target transformer in real time. The status data includes oil chromatography data, partial discharge signal, grounding current, local temperature data, and local vibration signal; wherein, the local temperature data includes local oil temperature and local shell temperature. The data processing module is used to preprocess the current state data, including: The current partial discharge signal is sequentially amplified, mixed, filtered, amplified, and filtered. Then, time-domain, frequency-domain, and time-frequency analysis are performed on a single discharge pulse of the partial discharge signal after the second filtering process to extract the time-domain, frequency-domain, and time-frequency features of the partial discharge signal. The current grounding current is sequentially subjected to third filtering, smoothing, data offset adjustment, and normalization. A multi-angle evaluation module is used to evaluate the state of the target transformer from multiple angles based on the preprocessed current state data, including: Based on the preprocessed local temperature data, assess whether the temperature rise and overall temperature of the target transformer are abnormal. Based on the preprocessed partial discharge signal, assess whether the insulation performance of the target transformer is abnormal; Based on the pre-processed local vibration signals, assess the winding loosening and core loosening of the target transformer; Based on the pre-processed grounding current, assess the grounding faults and insulation damage faults of the target transformer; Based on the pre-processed oil chromatography data, assess the overheating or discharge faults and fault locations of the target transformer. Based on the preprocessed current state data, identify the abnormal state of the target transformer; Based on the preprocessed current state data, a comprehensive fault diagnosis is performed on the target transformer; wherein the faults include partial discharge faults, mechanical faults, grounding faults, insulation faults, and overheating faults. The step of assessing whether the temperature rise of the target transformer is abnormal based on the preprocessed local temperature data includes: Based on multiple secondary influencing factors and current Using parameter combinations The first machine learning model predicts the current overall temperature of the target transformer. ;in, }, This indicates the quantity of the second influencing factor, where the second influencing factor refers to the factor that affects... Factors, This represents multiple local temperature data points of the target transformer; Calculate the target transformer in heating rate over time : ； in, This represents the overall temperature of the target transformer as predicted in the previous analysis. , Indicates the current The time point of collection, Indicates the previous The time point of data collection; like If the temperature rise rate exceeds the preset threshold, it indicates that the temperature rise of the target transformer is abnormal. The step of assessing whether the overall temperature of the target transformer is abnormal based on the preprocessed local temperature data includes: Based on multiple secondary influencing factors and current Based on the Bayesian model, it is determined according to the following formula. : = ； in, Indicates greater than of and several secondary influencing factors The following observations Likelihood, Indicates greater than of and several secondary influencing factors The prior probability, express Marginal likelihood, Indicates greater than of and several secondary influencing factors The posterior probability; where, The set temperature threshold; like If the probability exceeds the set probability threshold, it indicates that the temperature of the target transformer is abnormal; The step of evaluating whether the insulation performance of the target transformer is abnormal based on the preprocessed partial discharge signal includes: The time-domain features, frequency-domain features, and time-frequency features of the extracted partial discharge signal are input into the constructed insulation fault assessment model, and the insulation fault assessment model outputs the corresponding insulation threat level. If the insulation threat level output by the insulation fault assessment model is greater than the preset insulation threat level threshold, it indicates that the insulation performance of the target transformer is abnormal. The assessment of winding loosening and core loosening of the target transformer based on the preprocessed local vibration signal includes: If at least some of the current local vibration signals have frequencies that are integer multiples of 50Hz and amplitudes that are greater than a preset first amplitude threshold, it indicates that the windings of the target transformer are loose. If, among the current plurality of local vibration signals, at least some of the local vibration signals have a frequency greater than 20Hz and less than 50Hz and an amplitude greater than a preset second amplitude threshold, it indicates that the core of the target transformer is loose. The step of identifying the abnormal state of the target transformer based on the preprocessed current state data includes: Calculate the correlation coefficients between the current state data after preprocessing; Construct a correlation coefficient matrix based on the correlation coefficients between the data of each state; Based on the correlation coefficient matrix, an undirected graph network is constructed, wherein the name of each node in the undirected graph network is the name of each state data, the attribute value of each node is the corresponding state data value, and the edge is the connecting line between significantly related state data pairs; wherein, if the correlation coefficient of two state data is greater than a preset correlation coefficient threshold, they are defined as a group of significantly related state data pairs. Independent communities in the undirected graph network are identified using a community detection algorithm; Based on the identified independent communities, identify the abnormal state of the target transformer; The early warning module is used to issue an early warning for each abnormal state of the target transformer if the state of the target transformer is found to be abnormal from at least one angle.

Images