Fault early warning method and system for power transformer based on meteorological information

Abstract

The present invention relates to the technical field of power transformers, and in particular, provides a fault warning method and system for power transformers based on meteorological information. The method includes determining multi-source data according to warning indicators; cleaning and normalizing the multi-source data; fusing the processed multi-source data in time and space to generate a comprehensive data set containing multi-dimensional features, and extracting features highly related to transformer failures from the fused data set; constructing a multivariate coupling model to predict the probability of failure; setting different levels of warning thresholds according to the probability of failure output by the multivariate coupling model; and performing risk monitoring and warning release, and automatically notifying operation and maintenance personnel. The method can identify potential risks in advance, provide timely and effective maintenance decision support, ensure the safe and stable operation of the power grid, especially under extreme climatic conditions, and reduce the systemic risks caused by equipment failures.

Description

Technical Field

[0001] The present invention relates to the technical field of power transformers, and in particular to a fault early warning method and system for power transformers based on meteorological information. Background Art

[0002] With the continuous construction of new power systems based on new energy, all aspects of the power system, including power sources, transmission networks, and loads, are increasingly dependent on meteorological information. This dependence is not only reflected in the needs of power system operation, but also in the fact that meteorological conditions directly affect the safety and stability of the power system.

[0003] In the power supply link, photovoltaic power generation and wind power generation are connected to the power grid on a large scale and at a high proportion. Photovoltaic power generation is directly affected by the intensity of sunshine and cloud cover, while wind power generation is closely related to wind speed and wind direction, which leads to the stability of its output being directly affected by meteorological factors, and the grid dispatching also needs to be flexibly adjusted with the meteorological changes. Secondly, in the transmission network link, most of the power transmission and transformation equipment is exposed to the natural environment, high temperature, low temperature, humidity and other factors, thunderstorms, freezing and other extreme weather may have a direct impact on the mechanical properties, insulation properties and heat dissipation capacity of the equipment. For example, high temperature may accelerate the aging of equipment and reduce the insulation performance of transformer oil, while thunderstorms may cause overvoltage damage to the equipment. In terms of load, extreme weather such as cold waves and heat waves often cause a surge in electricity load, especially in terms of cooling and heating demand.

[0004] As a key component of the power transmission and transformation network of the power system, the operating environment of the transformer is also affected by meteorological factors. As the power grid energy changes from the traditional one-way flow to a more complex two-way flow, the operation of the transformer is affected not only by the power supply side, but also by the dual effects of the load side. Summary of the invention

[0005] In view of this, the present invention provides a fault warning method and system for power transformers based on meteorological information, which is used to identify potential risks in advance, provide timely and effective maintenance decision support, ensure the safe and stable operation of the power grid, especially under extreme climatic conditions, and reduce the systemic risks caused by equipment failures.

[0006] In a first aspect, the present invention provides a fault warning method for a power transformer based on meteorological information, the method comprising:

[0007] Step 1: Determine the data collection object according to the functional requirements, and plan the warning target, coverage and warning level according to the data resources and budget, and determine the warning indicators;

[0008] Step 2: Determine multi-source data according to the early warning indicators, where the multi-source data includes meteorological data, power supply side output power data, load side power consumption data, and transformer operation data;

[0009] Step 3: Clean and normalize the meteorological data, power supply side output power data, load side power consumption data, and transformer operation data to obtain processed data;

[0010] Step 4: Fuse the processed meteorological data, power supply side output power data, load side power consumption data, and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extract features that are highly correlated with transformer faults from the fused data set;

[0011] Step 5: Based on the deep learning algorithm, a multivariate coupling model is constructed that takes into account meteorological conditions, power supply side fluctuations, and load side fluctuations, and the comprehensive impact of multivariate factors on the operating status of the transformer is analyzed and predicted; the failure risk of the transformer is dynamically evaluated based on real-time input data; the internal stress and temperature of the transformer are obtained by analyzing meteorological fluctuations, power fluctuations, and load fluctuations, and the probability of failure is predicted;

[0012] Step 6: According to the probability of failure output by the multivariable coupling model, set different levels of warning thresholds, corresponding to slight risk, medium risk and high risk respectively; when the warning threshold is reached, the warning information is automatically generated, and the risk monitoring and warning release are carried out through the display of the control center, and the operation and maintenance personnel are automatically notified.

[0013] Optionally, collect meteorological data: obtain meteorological data related to transformer operation in real time, including temperature data, humidity data, wind speed data, thunderstorm data, precipitation data, and freezing data;

[0014] Collect output power data on the power supply side and power consumption data on the load side: obtain output power data on the power supply side and power demand data on the load side; the power supply side includes photovoltaic power generation or wind power generation;

[0015] Collect transformer operation data: Through the sensor network installed on the transformer, key operating parameters are monitored in real time, including winding temperature data, oil temperature data, transformer tap position data, load current data, voltage data, and vibration data.

[0016] Optionally, the step 3 includes:

[0017] Data cleaning: For missing data caused by sensor failure or communication interruption, use the mean, median, mode to fill in, or use interpolation to fill missing values ​​in time series data, where interpolation methods include linear interpolation; for outliers caused by data input errors or system failures, use the Z-Score method to detect outliers, and its expression is:

[0018] ;

[0019] If |Z|>3, then x is an outlier; μ is the sample mean, σ is the sample standard deviation;

[0020] Data normalization processing: It is used to convert data of different scales to the same scale, compress the data to a specific range of [0,1], and process it using the minimum and maximum normalization formula, which is expressed as:

[0021] ;

[0022] in, is the normalized data, x min and x max are the minimum and maximum values ​​in the data set, respectively.

[0023] Optionally, step 4 includes:

[0024] a. Data alignment: Align data from different sources to a unified time axis or spatial dimension; for time series data, resample the data to the same time interval to ensure the consistency of the data time, using the time linear interpolation formula, which is expressed as:

[0025] ;

[0026] in, For time t The interpolated value at time, t 1, t 2 is the time of adjacent known data points, y 1, y 2 is the value of the adjacent known data point;

[0027] b. Data fusion: The processed multi-source data is weighted and fused at the decision layer to generate a comprehensive data set containing multi-dimensional features. The expression of weighted fusion is:

[0028] ;

[0029] in, w i is the weight of the i-th data source, where i=1, 2, ..., n, F i is the characteristic of the i-th data source;

[0030] c. Feature extraction: It involves multiple levels of processing, including statistical features, frequency domain features and time features; statistical features include the mean, variance, kurtosis and autocorrelation coefficient of the data; frequency domain features include Fourier transform, fast Fourier transform and wavelet transform to show the periodicity and frequency characteristics of the data; time features include periodicity and trend characteristics to analyze the impact of long-term meteorological factors on transformer operation;

[0031] d. Feature selection: Use the Pearson correlation coefficient method to select features that are highly correlated with transformer faults to improve the prediction performance of the model. The expression is:

[0032] ;

[0033] in, Cov ( X , Y ) is X and Y The covariance of σ X and σ Y is its standard deviation.

[0034] Optionally, step 5 includes:

[0035] Using the extracted feature data, a deep learning model, long short-term memory network (LSTM), is used to build a fault prediction model. The historical operating data and meteorological conditions of the transformer are used to predict whether the transformer will fail within the next 120 minutes. The model architecture is designed as the input of the operating parameters and meteorological data of the transformer, and the LSTM network is set to 2 layers, with 50 units in each layer, a regularization Dropout value of 0.2, and an output layer with a sigmoid-activated single neuron, which outputs the probability of fault occurrence. The evaluation indicators include ROC-AUC value, accuracy, and recall rate.

[0036] Optionally, step 6 includes:

[0037] Based on the predicted probability of failure, an initial warning threshold is set. If the predicted probability of failure is greater than the initial warning threshold, an alarm is triggered.

[0038] Set different levels of warning thresholds, including:

[0039] Warning: When the predicted failure probability is less than the warning threshold, a level 1 alarm is triggered to notify the operation and maintenance personnel to conduct a preliminary inspection;

[0040] Severe alarm: When the predicted failure probability is greater than or equal to the warning threshold, a level 2 alarm is triggered and preventive measures are taken immediately;

[0041] Emergency alarm: When the predicted failure probability is greater than the warning threshold and the trend continues to rise, a level 3 alarm is triggered, requiring emergency shutdown or isolation of the equipment.

[0042] In a second aspect, the present invention provides a fault warning system for a power transformer based on meteorological information, the system comprising:

[0043] The demand analysis module is used to determine the objects of data collection according to functional requirements, and to plan the warning targets, coverage and warning levels according to data resources and budget, and determine the warning indicators;

[0044] A data acquisition module is used to determine multi-source data according to the early warning indicators, wherein the multi-source data includes meteorological data, power output data on the power supply side, power consumption data on the load side, and transformer operation data;

[0045] The data processing module is used to clean and normalize the meteorological data, the output power data on the power supply side, the power consumption data on the load side, and the transformer operation data to obtain the processed data;

[0046] The data fusion and feature extraction module is used to fuse the processed meteorological data, power output data on the power supply side, power consumption data on the load side, and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extract features that are highly correlated with transformer faults from the fused data set;

[0047] The fault prediction model building module is used to build a multivariate coupling model based on deep learning algorithms that takes into account meteorological conditions, power supply side fluctuations, and load side fluctuations, and analyze and predict the comprehensive impact of multivariate factors on the operating status of the transformer; dynamically evaluate the fault risk of the transformer based on real-time input data; obtain the internal stress and temperature of the transformer by analyzing meteorological fluctuations, power fluctuations, and load fluctuations, and predict the probability of fault occurrence;

[0048] The intelligent early warning module is used to set different levels of early warning thresholds according to the probability of failure output by the multivariable coupling model, corresponding to slight risk, medium risk and high risk respectively; when the early warning threshold is reached, early warning information is automatically generated, and risk monitoring and early warning are issued through the display of the control center, and the operation and maintenance personnel are automatically notified.

[0049] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is running, the electronic device where the computer-readable storage medium is located is controlled to execute the fault warning method for power transformers based on meteorological information in the first aspect or any possible implementation of the first aspect.

[0050] In a fourth aspect, an embodiment of the present invention provides an electronic device, comprising: one or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions, which, when executed by the electronic device, enable the device to execute the fault warning method for power transformers based on meteorological information in the first aspect or any possible implementation of the first aspect.

[0051] In the technical solution provided by the present invention, the method includes determining the object of data collection according to functional requirements, and planning the warning target, coverage and warning level according to data resources and budget, and determining the warning indicators; determining multi-source data according to the warning indicators, wherein the multi-source data includes meteorological data, power supply side output power data, load side power consumption data and transformer operation data; cleaning and normalizing the meteorological data, power supply side output power data, load side power consumption data and transformer operation data to obtain processed data; fusing the processed meteorological data, power supply side output power data, load side power consumption data and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extracting features highly correlated with transformer faults from the fused data set; constructing a deep learning algorithm based on meteorological data. A multivariate coupling model of meteorological conditions, power supply side fluctuations and load side fluctuations is constructed, and the comprehensive impact of multivariate factors on the operating status of the transformer is analyzed and predicted; the failure risk of the transformer is dynamically evaluated based on real-time input data; the internal stress and temperature of the transformer are obtained by analyzing meteorological fluctuations, power fluctuations and load fluctuations, and the probability of failure is predicted; according to the probability of failure output by the multivariate coupling model, different levels of early warning thresholds are set, corresponding to slight risk, medium risk and high risk respectively; when the early warning threshold is reached, early warning information is automatically generated, and risk monitoring and early warning release are carried out through the display of the control center, and the operation and maintenance personnel are automatically notified. This method can identify potential risks in advance, provide timely and effective maintenance decision support, ensure the safe and stable operation of the power grid, especially under extreme climatic conditions, and reduce the systemic risks caused by equipment failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1 A flowchart of a fault warning method provided by an embodiment of the present invention;

[0054] Figure 2 A schematic diagram of a fault warning system provided by an embodiment of the present invention;

[0055] Figure 3 A schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] It should be clear that the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention are also intended to include plural forms, unless the context clearly indicates other meanings.

[0059] It should be understood that the term "and / or" used in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0060] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0061] Figure 1 A flowchart of a fault warning method provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes:

[0062] Step 1: Determine the data collection object according to functional requirements, and plan the warning target, coverage and warning level according to data resources and budget, and determine the warning indicators.

[0063] In the embodiment of the present invention, the functional requirements of the system are clarified, the objects of data collection are determined, the core functions of the system are specified, a priority analysis is performed, and the system's early warning targets, coverage, early warning levels and other indicators are planned in combination with existing technical conditions, data resources and budgets.

[0064] Step 2: Determine multi-source data based on early warning indicators, where the multi-source data includes meteorological data, power supply side output power data, load side power consumption data, and transformer operation data.

[0065] In the embodiment of the present invention, necessary sensors, data acquisition devices, sensor networks and communication networks are installed to ensure the real-time and accuracy of data acquisition.

[0066] In the embodiment of the present invention, step 2 includes:

[0067] Collect meteorological data: obtain meteorological data related to transformer operation in real time, including temperature data, humidity data, wind speed data, thunderstorm data, precipitation data, and freezing data;

[0068] In the embodiment of the present invention, the meteorological factors that may affect the power output and load demand are mainly monitored.

[0069] Collect output power data on the power supply side and power consumption data on the load side: obtain output power data on the power supply side and power demand data on the load side; the power supply side includes photovoltaic power generation or wind power generation;

[0070] Collect transformer operation data: Through the sensor network installed on the transformer, key operating parameters are monitored in real time, including winding temperature data, oil temperature data, transformer tap position data, load current data, voltage data, and vibration data.

[0071] Step 3: Clean and normalize the meteorological data, power supply side output power data, load side power consumption data, and transformer operation data to obtain processed data.

[0072] In the embodiment of the present invention, the data is cleaned and normalized to eliminate abnormal data, ensure the consistency of the data, and provide high-quality data for subsequent analysis and modeling.

[0073] In the embodiment of the present invention, step 3 includes:

[0074] Data cleaning: For missing data caused by sensor failure or communication interruption, use the mean, median, mode to fill in, or use interpolation to fill missing values ​​in time series data, where interpolation methods include linear interpolation; for outliers caused by data input errors or system failures, use the Z-Score method to detect outliers, and its expression is:

[0075] ;

[0076] If |Z|>3, then x is an outlier; μ is the sample mean, σ is the sample standard deviation;

[0077] Data normalization processing: It is used to convert data of different scales to the same scale, compress the data to a specific range of [0,1], and process it using the minimum and maximum normalization formula, which is expressed as:

[0078] ;

[0079] in, is the normalized data, x min and x max are the minimum and maximum values ​​in the data set, respectively.

[0080] In the embodiment of the present invention, data normalization processing can prevent certain features from having an unbalanced impact on model training.

[0081] Step 4: Fuse the processed meteorological data, power supply side output power data, load side power consumption data, and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extract features that are highly correlated with transformer failures from the fused data set.

[0082] In the embodiment of the present invention, step 4 includes:

[0083] Data from different sources are processed comprehensively to obtain more comprehensive information, and then key features that are helpful for model prediction are extracted from these fused data.

[0084] a. Data alignment: Align data from different sources to a unified time axis or spatial dimension; for time series data, resample the data to the same time interval to ensure the consistency of the data time, using the time linear interpolation formula, which is expressed as:

[0085] ;

[0086] in, For timet The interpolated value at time, t 1, t 2 is the time of adjacent known data points, y 1, y 2 is the value of the adjacent known data point;

[0087] For spatial data, mapping data from different geographic locations to the same coordinate system ensures geographic consistency.

[0088] b. Data fusion: The processed multi-source data is weighted and fused at the decision layer to generate a comprehensive data set containing multi-dimensional features. The expression of weighted fusion is:

[0089] ;

[0090] in, w i is the weight of the i-th data source, where i=1, 2, ..., n, F i is the characteristic of the i-th data source;

[0091] c. Feature extraction: It involves multiple levels of processing, including statistical features, frequency domain features and time features; statistical features include the mean, variance, kurtosis and autocorrelation coefficient of the data; frequency domain features include Fourier transform, fast Fourier transform and wavelet transform to show the periodicity and frequency characteristics of the data; time features include periodicity and trend characteristics to analyze the impact of long-term meteorological factors on transformer operation;

[0092] d. Feature selection: Use the Pearson correlation coefficient method to select features that are highly correlated with transformer faults to improve the prediction performance of the model. The expression is:

[0093] ;

[0094] in, Cov ( X , Y ) is X and Y The covariance of σ X and σ Y is its standard deviation.

[0095] Step 5. Based on the deep learning algorithm, a multivariable coupling model is constructed that takes into account meteorological conditions, power supply side fluctuations, and load side fluctuations, and the comprehensive impact of multivariable factors on the operating status of the transformer is analyzed and predicted; the failure risk of the transformer is dynamically evaluated based on real-time input data; the internal stress and temperature of the transformer are obtained by analyzing meteorological fluctuations, power fluctuations, and load fluctuations, and the probability of failure is predicted.

[0096] The Long Short-Term Memory (LSTM) network is a special recursive neural network that solves the problem that traditional recursive neural networks are prone to losing long-term dependent information in long sequences. It controls the flow of information by introducing a memory unit and three gating mechanisms, namely the input gate, forget gate, and output gate, thereby effectively memorizing and updating important information over a long period of time.

[0097] The memory cell is the core part of the long short-term memory network. It is used to store long-term dependency information from the remote end to the current time step, and determines whether to retain or discard the information through various gating mechanisms. The input gate determines how much input information can be updated to the memory cell, the forget gate determines the information that should be discarded in the memory cell, and the output gate determines the hidden state of the output.

[0098] The model construction of the long short-term memory network usually consists of an input layer, a hidden layer, and an output layer. The input layer receives a multidimensional time series. The input shape is usually the number of samples for each training, the number of time steps for each sample, and the number of data features for each time step (such as the number of temperatures, currents, etc.); the hidden layer is used to find time dependencies. Its design needs to consider the following factors: the number of layers, the number of hidden units, and Dropout regularization. The number of layers determines the ability of the long short-term memory network to capture time relationships. The more complex the time dependency, the more layers are required. The number of hidden units in each long short-term memory network layer determines the capacity of the model, which needs to be adjusted according to the size and complexity of the data. Add a Dropout layer to the hidden layer to prevent overfitting. The common Dropout value is 0.2 to 0.5; the design of the output layer depends on the needs of the specific task. For prediction or not problems, a single neuron is used, and the sigmoid activation function is used to output the predicted fault probability.

[0099] In the embodiment of the present invention, step 5 includes:

[0100] Using the extracted feature data, a deep learning model, long short-term memory network (LSTM), is used to build a fault prediction model. The historical operating data and meteorological conditions of the transformer are used to predict whether the transformer will fail within the next 120 minutes. The model architecture is designed as the input of the operating parameters and meteorological data of the transformer, and the LSTM network is set to 2 layers, with 50 units in each layer, a regularization Dropout value of 0.2, and an output layer with a sigmoid-activated single neuron, which outputs the probability of fault occurrence. The evaluation indicators include ROC-AUC value, accuracy, and recall rate.

[0101] In an embodiment of the present invention, if the model performance does not meet the preset standard, the prediction accuracy is improved by increasing the number of LSTM network layers, adjusting the learning rate or optimizing the input features; when the probability of a fault occurring is greater than a predetermined threshold, an early warning is triggered so that protective measures can be taken in advance.

[0102] Step 6: According to the probability of failure output by the multivariable coupling model, set different levels of warning thresholds, corresponding to slight risk, medium risk and high risk respectively; when the warning threshold is reached, the warning information is automatically generated, and the risk monitoring and warning release are carried out through the display of the control center, and the operation and maintenance personnel are automatically notified.

[0103] In the embodiment of the present invention, step 6 includes:

[0104] Based on the predicted probability of failure, an initial warning threshold is set. If the predicted probability of failure is greater than the initial warning threshold, an alarm is triggered.

[0105] Set different levels of warning thresholds, including:

[0106] Warning: When the predicted failure probability is less than the warning threshold, a level 1 alarm is triggered to notify the operation and maintenance personnel to conduct a preliminary inspection;

[0107] Severe alarm: When the predicted failure probability is greater than or equal to the warning threshold, a level 2 alarm is triggered and preventive measures are taken immediately;

[0108] Emergency alarm: When the predicted failure probability is greater than the warning threshold and the trend continues to rise, a level 3 alarm is triggered, requiring emergency shutdown or isolation of the equipment.

[0109] In a possible implementation, the initial warning threshold is set to 0.7. When the predicted failure probability is in the range of [0.6-0.7), when the predicted failure probability is less than the warning threshold, a level one alarm is triggered and the operation and maintenance personnel are notified to conduct a preliminary inspection; when the predicted failure probability is in the range of [0.7-0.9], when the predicted failure probability is greater than or equal to the warning threshold, a level two alarm is triggered and preventive measures are taken immediately; when the predicted failure probability is greater than 0.9, when the predicted failure probability is greater than the warning threshold and the trend continues to rise, a level three alarm is triggered, requiring emergency shutdown or isolation of the equipment.

[0110] Figure 2 A schematic diagram of a fault warning system provided by an embodiment of the present invention is shown in FIG. Figure 2 As shown, the system includes: a demand analysis module 1, a data acquisition module 2, a data processing module 3, a data fusion and feature extraction module 4, a fault prediction model construction module 5 and an intelligent early warning module 6; the demand analysis module 1 is connected to the data acquisition module 2, the data acquisition module 2 is connected to the data processing module 3, the data processing module 3 is connected to the data fusion and feature extraction module 4, the data fusion and feature extraction module 4 is connected to the fault prediction model construction module 5, and the fault prediction model construction module 5 is connected to the intelligent early warning module 6.

[0111] Demand analysis module 1 is used to determine the object of data collection according to functional requirements, and to plan the warning target, coverage and warning level according to data resources and budget, and to determine the warning indicators; data collection module 2 is used to determine multi-source data according to the warning indicators, where the multi-source data includes meteorological data, power supply side output power data, load side power consumption data and transformer operation data; data processing module 3 is used to clean and normalize the meteorological data, power supply side output power data, load side power consumption data and transformer operation data to obtain processed data; data fusion and feature extraction module 4 is used to fuse the processed meteorological data, power supply side output power data, load side power consumption data and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extract the data from the fusion The features highly correlated with transformer failures are extracted from the data set after the failure prediction. The fault prediction model construction module 5 is used to construct a multivariate coupling model that takes into account meteorological conditions, power supply side fluctuations and load side fluctuations based on a deep learning algorithm, and analyze and predict the comprehensive impact of multivariate factors on the operating status of the transformer. The failure risk of the transformer is dynamically evaluated according to the real-time input data. The internal stress and temperature of the transformer are obtained by analyzing meteorological fluctuations, power fluctuations and load fluctuations, and the probability of failure is predicted. The intelligent early warning module 6 is used to set different levels of early warning thresholds according to the probability of failure output by the multivariate coupling model, corresponding to slight risk, medium risk and high risk respectively. When the early warning threshold is reached, the early warning information is automatically generated, and the risk monitoring and early warning are issued through the display of the control center, and the operation and maintenance personnel are automatically notified.

[0112] In the embodiment of the present invention, the trained fault prediction long short-term memory network model is encapsulated into a callable API interface, which provides the function of predicting the probability of faults and can process real-time input data. The fault warning system is integrated into the power dispatching and operation and maintenance system to ensure that the operation and maintenance personnel can directly access the prediction results through the system interface, and realize automated risk monitoring and warning release.

[0113] Each step of the embodiment of the present invention may be performed by an electronic device, which includes but is not limited to a mobile phone, a tablet computer, a portable PC, a desktop computer, etc.

[0114] In the technical solution provided by the present invention, the method includes determining the object of data collection according to functional requirements, and planning the warning target, coverage and warning level according to data resources and budget, and determining the warning indicators; determining multi-source data according to the warning indicators, wherein the multi-source data includes meteorological data, power supply side output power data, load side power consumption data and transformer operation data; cleaning and normalizing the meteorological data, power supply side output power data, load side power consumption data and transformer operation data to obtain processed data; fusing the processed meteorological data, power supply side output power data, load side power consumption data and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extracting features highly correlated with transformer faults from the fused data set; constructing a deep learning algorithm based on meteorological data. A multivariate coupling model of meteorological conditions, power supply side fluctuations and load side fluctuations is constructed, and the comprehensive impact of multivariate factors on the operating status of the transformer is analyzed and predicted; the failure risk of the transformer is dynamically evaluated based on real-time input data; the internal stress and temperature of the transformer are obtained by analyzing meteorological fluctuations, power fluctuations and load fluctuations, and the probability of failure is predicted; according to the probability of failure output by the multivariate coupling model, different levels of early warning thresholds are set, corresponding to slight risk, medium risk and high risk respectively; when the early warning threshold is reached, early warning information is automatically generated, and risk monitoring and early warning release are carried out through the display of the control center, and the operation and maintenance personnel are automatically notified. This method can identify potential risks in advance, provide timely and effective maintenance decision support, ensure the safe and stable operation of the power grid, especially under extreme climatic conditions, and reduce the systemic risks caused by equipment failure.

[0115] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the electronic device where the computer-readable storage medium is located is controlled to execute the above-mentioned embodiment of the fault warning method for power transformer based on meteorological information.

[0116] Figure 3 A schematic diagram of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, the electronic device 21 includes: a processor 211, a memory 212, and a computer program 213 stored in the memory 212 and executable on the processor 211. When the computer program 213 is executed by the processor 211, the fault warning method for the power transformer based on meteorological information in the embodiment is implemented. To avoid repetition, they are not described one by one here.

[0117] The electronic device 21 includes, but is not limited to, a processor 211 and a memory 212. Those skilled in the art will appreciate that the figure is merely an example of the electronic device 21 and does not constitute a limitation on the electronic device 21, which may include more or fewer components than shown in the figure, or a combination of certain components, or different components, for example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0118] The processor 211 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0119] The memory 212 may be an internal storage unit of the electronic device 21, such as a hard disk or memory of the electronic device 21. The memory 212 may also be an external storage device of the electronic device 21, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card (FlashCard), etc. equipped on the electronic device 21. Further, the memory 212 may also include both an internal storage unit of the electronic device 21 and an external storage device. The memory 212 is used to store computer programs and other programs and data required by network devices. The memory 212 may also be used to temporarily store data that has been output or is to be output.

[0120] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0121] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A fault warning method for power transformer based on meteorological information, characterized in that: The method comprises: Step 1: Determine the data collection object according to the functional requirements, and plan the warning target, coverage and warning level according to the data resources and budget, and determine the warning indicators; Step 2: Determine multi-source data according to the early warning indicators, where the multi-source data includes meteorological data, power supply side output power data, load side power consumption data, and transformer operation data; Step 3: Clean and normalize the meteorological data, power supply side output power data, load side power consumption data, and transformer operation data to obtain processed data; Step 4: Fuse the processed meteorological data, power supply side output power data, load side power consumption data, and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extract features that are highly correlated with transformer faults from the fused data set; Step 5: Based on the deep learning algorithm, a multivariate coupling model is constructed that takes into account meteorological conditions, power supply side fluctuations, and load side fluctuations, and the comprehensive impact of multivariate factors on the operating status of the transformer is analyzed and predicted; the failure risk of the transformer is dynamically evaluated based on real-time input data; the internal stress and temperature of the transformer are obtained by analyzing meteorological fluctuations, power fluctuations, and load fluctuations, and the probability of failure is predicted; Step 6: According to the probability of failure output by the multivariable coupling model, set different levels of warning thresholds, corresponding to slight risk, medium risk and high risk respectively; when the warning threshold is reached, the warning information is automatically generated, and the risk monitoring and warning release are carried out through the display of the control center, and the operation and maintenance personnel are automatically notified.

2. The method according to claim 1, characterized in that The step 2 comprises: Collect meteorological data: obtain meteorological data related to transformer operation in real time, including temperature data, humidity data, wind speed data, thunderstorm data, precipitation data, and freezing data; Collect output power data on the power supply side and power consumption data on the load side: obtain output power data on the power supply side and power demand data on the load side; the power supply side includes photovoltaic power generation or wind power generation; Collect transformer operation data: Through the sensor network installed on the transformer, key operating parameters are monitored in real time, including winding temperature data, oil temperature data, transformer tap position data, load current data, voltage data, and vibration data.

3. The method according to claim 1, characterized in that The step 3 comprises: Data cleaning: For missing data caused by sensor failure or communication interruption, use the mean, median, mode to fill in, or use interpolation to fill missing values ​​in time series data, where interpolation methods include linear interpolation; for outliers caused by data input errors or system failures, use the Z-Score method to detect outliers, and its expression is: ； If |Z|>3, then x is an outlier; μ is the sample mean, σ is the sample standard deviation; Data normalization processing: It is used to convert data of different scales to the same scale, compress the data to a specific range of [0, 1], and use the minimum and maximum normalization formula for processing. The expression is: ； in, is the normalized data, x min and x max are the minimum and maximum values ​​in the data set, respectively.

4. The method according to claim 1, characterized in that: The step 4 comprises: a. Data alignment: Align data from different sources to a unified time axis or spatial dimension; for time series data, resample the data to the same time interval to ensure the consistency of the data time, using the time linear interpolation formula, which is expressed as: ； in, For time t The interpolated value at time, t 1, t 2 is the time of adjacent known data points, y 1, y 2 is the value of the adjacent known data point; b. Data fusion: The processed multi-source data is weighted and fused at the decision layer to generate a comprehensive data set containing multi-dimensional features. The expression of weighted fusion is: ； in, w i is the weight of the i-th data source, where i=1, 2, ..., n, F i is the characteristic of the i-th data source; c. Feature extraction: It involves multiple levels of processing, including statistical features, frequency domain features and time features; statistical features include the mean, variance, kurtosis and autocorrelation coefficient of the data; frequency domain features include Fourier transform, fast Fourier transform and wavelet transform to show the periodicity and frequency characteristics of the data; time features include periodicity and trend characteristics to analyze the impact of long-term meteorological factors on transformer operation; d. Feature selection: Use the Pearson correlation coefficient method to select features that are highly correlated with transformer faults to improve the prediction performance of the model. The expression is: ； in, Cov ( X , Y ) is X and Y The covariance of σ X and σ Y is its standard deviation.

5. The method according to claim 1, characterized in that The step 5 comprises: Using the extracted feature data, a deep learning model, long short-term memory network (LSTM), is used to build a fault prediction model. The historical operating data and meteorological conditions of the transformer are used to predict whether the transformer will fail within the next 120 minutes. The model architecture is designed as the input of the operating parameters and meteorological data of the transformer, and the LSTM network is set to 2 layers, with 50 units in each layer, a regularization Dropout value of 0.2, and an output layer with a sigmoid-activated single neuron, which outputs the probability of fault occurrence. The evaluation indicators include ROC-AUC value, accuracy, and recall rate.

6. The method according to claim 1, characterized in that The step 6 comprises: Based on the predicted probability of failure, an initial warning threshold is set. If the predicted probability of failure is greater than the initial warning threshold, an alarm is triggered. Set different levels of warning thresholds, including: Warning: When the predicted failure probability is less than the warning threshold, a level 1 alarm is triggered to notify the operation and maintenance personnel to conduct a preliminary inspection; Severe alarm: When the predicted failure probability is greater than or equal to the warning threshold, a level 2 alarm is triggered and preventive measures are taken immediately; Emergency alarm: When the predicted failure probability is greater than the warning threshold and the trend continues to rise, a level 3 alarm is triggered, requiring emergency shutdown or isolation of the equipment.

7. A fault warning system for power transformers based on meteorological information, characterized in that: The system comprises: The demand analysis module is used to determine the objects of data collection according to functional requirements, and to plan the warning targets, coverage and warning levels according to data resources and budget, and determine the warning indicators; A data acquisition module is used to determine multi-source data according to the early warning indicators, wherein the multi-source data includes meteorological data, power output data on the power supply side, power consumption data on the load side, and transformer operation data; The data processing module is used to clean and normalize the meteorological data, the output power data on the power supply side, the power consumption data on the load side, and the transformer operation data to obtain the processed data; The data fusion and feature extraction module is used to fuse the processed meteorological data, power output data on the power supply side, power consumption data on the load side, and transformer operation data in time and space to generate a comprehensive data set containing multi-dimensional features, and extract features that are highly correlated with transformer faults from the fused data set; The fault prediction model building module is used to build a multivariate coupling model based on deep learning algorithms that takes into account meteorological conditions, power supply side fluctuations, and load side fluctuations, and analyze and predict the comprehensive impact of multivariate factors on the operating status of the transformer; dynamically evaluate the fault risk of the transformer based on real-time input data; obtain the internal stress and temperature of the transformer by analyzing meteorological fluctuations, power fluctuations, and load fluctuations, and predict the probability of fault occurrence; The intelligent early warning module is used to set different levels of early warning thresholds according to the probability of failure output by the multivariable coupling model, corresponding to slight risk, medium risk and high risk respectively; when the early warning threshold is reached, early warning information is automatically generated, and risk monitoring and early warning are issued through the display of the control center, and the operation and maintenance personnel are automatically notified.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is running, the computer-readable storage medium is controlled to execute the fault warning method for power transformer based on meteorological information as described in any one of claims 1 to 6.

9. An electronic device, characterized in that: include: one or more processors; Memory; And one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions, which, when executed by the electronic device, enable the device to execute the fault warning method for power transformer based on meteorological information as described in any one of claims 1 to 6.

Images