Power transformer state evaluation method based on multi-scale gating attention all convolutional network

Abstract

The present application relates to a power transformer state evaluation method based on a multi-scale gated attention all convolutional network, which solves the problem of lack of reliable and comprehensive power transformer state evaluation method in using multi-source sensor data compared with the prior art.The present application comprises the following steps: acquisition and preprocessing of power transformer multi-source sensor data; construction of power transformer state evaluation data set; construction of power transformer state evaluation model; training of power transformer state evaluation model; real-time monitoring of power transformer; evaluation of power transformer state.The present application realizes an end-to-end power transformer state evaluation model based on multi-source sensor data, which can comprehensively and real-timely evaluate the running state of the power transformer by using the time series data of the multi-source sensor, thereby improving the reliability of the transformer state evaluation.

Description

Technical Field

[0001] The present invention relates to the technical field of power transformer monitoring, and in particular to a power transformer state assessment method based on a multi-scale gated attention fully convolutional network. Background Art

[0002] Power transformers are a critical component of the power system. Unexpected failures can severely impact the stable operation of the power grid, resulting in significant losses for both power companies and users. Therefore, scientifically and accurately assessing the operating status of substation equipment is crucial for promptly identifying potential power outages and enabling proactive maintenance and repair of transformers.

[0003] In recent years, scholars have continuously explored methods for evaluating the condition of substation equipment, resulting in several mainstream approaches. For example, methods based on dissolved gas analysis (DGA) primarily analyze the composition and concentration of gases dissolved in transformer oil to assess transformer condition. Other methods include partial discharge (PD) detection technology, thermal analysis based on infrared thermal imaging or transformer oil temperature, vibration analysis, and frequency response analysis. However, these methods have limitations. Based solely on single data points, they cannot comprehensively assess transformer operating conditions and therefore fail to meet current needs.

[0004] To address the limitations of single-data evaluation, some researchers have explored comprehensive assessments of substation equipment status based on multiple factors. These methods consider the impact of multiple factors on substation equipment status evaluation during design. However, they also have certain limitations. Traditional methods struggle to ensure reliable predictions for complex multivariate data. Deep learning, on the other hand, has been shown to be able to learn more complex features and achieve higher prediction accuracy.

[0005] Furthermore, these evaluation methods rely heavily on current data and are unable to fully assess the transformer's operating status. As we all know, the transformer's operating status is not only related to current data, but also to data from past periods of time. In other words, a period of time series data can more accurately reflect the transformer's operating status.

[0006] Based on the limitations of the above work, we have two problems to solve: 1) comprehensively evaluate the operating status of the transformer based on multi-source sensor data; 2) construct the current data and historical data of the past period into a time series data, and evaluate the operating status of the transformer based on the time series data.

[0007] Therefore, based on the above two problems, how to develop transformer condition evaluation technology based on multivariate time series data has become a technical problem that needs to be solved urgently. Summary of the Invention

[0008] The purpose of the present invention is to solve the problem in the prior art of the lack of a reliable and comprehensive power transformer state assessment method using multi-source sensor data, and to provide a power transformer state assessment method based on a multi-scale gated attention fully convolutional network to solve the above problem.

[0009] In order to achieve the above object, the technical solution of the present invention is as follows:

[0010] A method for power transformer condition assessment based on a multi-scale gated attention fully convolutional network includes the following steps:

[0011] Acquisition and preprocessing of multi-source sensor data for power transformers: Acquisition and preprocessing of multi-source sensor data for power transformers, including dissolved gas in oil, top oil temperature, core grounding current, and power supply load data;

[0012] Construction of a power transformer condition evaluation dataset: Based on the preprocessed multi-source sensor data, a two-dimensional matrix dataset with multiple variables and labels with time series characteristics is constructed as the power transformer condition evaluation dataset;

[0013] Construction of power transformer condition assessment model: Constructing a power transformer condition assessment model based on a multi-scale gated attention module and a fully convolutional network module;

[0014] Training of power transformer condition assessment model: The pre-processed power transformer basic data is fed into the power transformer condition assessment model for training;

[0015] Real-time monitoring of power transformers: Real-time monitoring of power transformers to obtain monitoring data and pre-process it;

[0016] Evaluation of the power transformer status: The preprocessed real-time monitoring data is input into the trained power transformer status evaluation model to obtain the evaluation results of the power transformer status.

[0017] The construction of the power transformer condition evaluation data set includes the following steps:

[0018] The transformer's multi-source sensor data, including dissolved gas in oil, trace water, core grounding current, top oil temperature, and power supply load data, is acquired from the substation's data monitoring platform. This data is then merged by time to create a complete data table.

[0019] Preprocess the merged data table, manually analyze multiple variables separately, deal with missing values ​​and outliers for each variable, and then use the min-max normalization method to normalize the data to form a preprocessed data table;

[0020] The preprocessed data table is fused with the transformer operating status data according to the time dimension to construct a power transformer status evaluation dataset for training, where the last column is the label column and the remaining columns are feature columns.

[0021] The construction of the power transformer condition assessment model includes the following steps:

[0022] Construct a power transformer condition assessment model, which includes an input layer, a feature extraction layer, a feature fusion layer, and a classification layer;

[0023] Set up the input layer, which is used to read the raw data and reconstruct the data into the data format used by the model;

[0024] Set the feature extraction layer, which consists of a multi-scale gated attention module and a fully convolutional network module.

[0025] Set up a multi-scale gated attention module, which consists of multi-scale convolution, gating mechanism and multi-head self-attention mechanism.

[0026] Multi-scale convolution is set to extract multi-scale temporal information of different variables. It consists of three one-dimensional convolutions with different convolution kernel sizes. The sizes of the three one-dimensional convolution kernels are set to 1*3, 1*5, and 1*7 respectively. The features extracted by the three one-dimensional convolutions are then spliced ​​to obtain the output of the multi-scale convolution module.

[0027]

[0028] in, Represents the multi-scale features after splicing and fusion, is the transpose of the input data, DWC1d is a one-dimensional depth-wise separable convolution;

[0029] Set the gating mechanism, which consists of a learnable one-dimensional vector, a sigmoid function, and a bitwise multiplication operation, as shown below:

[0030] F gate =σ(Conv1d(F ms ))⊙F ms ,

[0031] in, represents the gated weighted output feature, σ represents the Sigmoid activation function, and ⊙ is the element-wise multiplication operation;

[0032] Set up a multi-head self-attention mechanism, which consists of feature splicing operations and multi-head attention operations.

[0033]

[0034] in, represents the input features of multi-head self-attention, is the linear transformation matrix, are the queries, keys, and values ​​required for self-attention calculation after linear transformation. is a matrix multiplication operation, and d is the variable embedding dimension;

[0035] The fully convolutional network module is set up, which consists of three layers of convolution. The first and third layers of convolution are ordinary one-dimensional convolutions, and the second layer is a one-dimensional depth-separable convolution.

[0036] X′ i =Conv1d(DWC1d(Conv1d(X i )))

[0037] f=LN((X′ i ⊙X i )+X i ),

[0038] Among them, X i is the input feature, X′ i It is the output feature after three layers of convolution, and LN represents layer normalization;

[0039] Set the feature fusion layer,

[0040] The feature fusion layer consists of vector concatenation and vector addition operations. It concatenates the output of the gating mechanism with the output of the multi-head self-attention mechanism. A feedforward neural network is set to reconstruct the concatenated features. The reconstructed output features are then added to the output of the fully convolutional network module to achieve feature fusion.

[0041] The classification layer is set, and the fused features are input into the MLP network for classification to obtain the status category of the power transformer, thereby realizing the status evaluation of the power transformer.

[0042] The training of the power transformer condition assessment model comprises the following steps:

[0043] The constructed power transformer condition evaluation dataset is divided into a training set and a test set, and then the training set and the test set are reconstructed into data formats readable by the model, and the training set data is input into the input layer of the power transformer condition evaluation model;

[0044] The feature extraction layer extracts effective features from the original data, and the multi-scale gated attention module and the fully convolutional network module are used to extract the temporal features of the data and the dependencies between variables.

[0045] Multi-scale gated attention module extracts temporal features:

[0046] Multi-scale convolution is used to extract multi-scale temporal information of multiple variables, and multi-scale features are obtained through feature concatenation. The multi-scale features are then input into a gating mechanism, which learns a gating vector. By multiplying the gating vector with the multi-scale features, the important temporal features of each variable are retained and redundant time scale information is filtered out.

[0047] The global temporal features of the original data are extracted through the multi-head self-attention mechanism, and the multi-scale features are fused with the original data features and input into the multi-head self-attention mechanism to learn and capture the global temporal features of the time series;

[0048] The fully convolutional network module extracts dependencies between variables:

[0049] The original data is input into the fully convolutional network to extract the interactive dependencies between multiple variables;

[0050] The feature fusion layer fuses multiple extracted valid features to obtain comprehensive features:

[0051] The output features of the gating mechanism are fused with the output features of the multi-head self-attention mechanism to obtain the time features, and the time features are fused with the features output by the full convolutional network module to obtain the omnidirectional features.

[0052] The classification layer performs classification predictions:

[0053] The extracted all-round features are used for classification prediction. The predicted value and the true value are calculated through the cross-entropy loss function to calculate the loss value. Then, the model parameters are updated through Adam and back propagation to continuously reduce the loss value until it stabilizes. The power transformer condition evaluation model is obtained. Subsequently, the model performance is evaluated using the test set. If the model performance meets the preset requirements, it is applied to the power transformer condition evaluation. Otherwise, the model hyperparameters need to be adjusted according to the evaluation results, and the model is retrained using the training set data until a power transformer condition evaluation model that meets the performance requirements is obtained.

[0054] When the time series data from multi-source sensors are input into the power transformer status evaluation model, the model will evaluate the current transformer status based on the data characteristics.

[0055] Beneficial effects

[0056] Compared with the prior art, the power transformer state assessment method based on a multi-scale gated attention fully convolutional network of the present invention realizes an end-to-end power transformer state assessment model based on data from multi-source sensors. The model can use the time series data of multi-source sensors to perform a comprehensive real-time assessment of the operating state of the power transformer, thereby improving the reliability of the transformer state assessment.

[0057] Experiments show that the proposed method achieves state-of-the-art performance in terms of accuracy for identifying substation status. Furthermore, extensive experiments on publicly available multivariate time series datasets validate the generalizability of the proposed method. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a method sequence diagram of the present invention;

[0059] Figure 2 This is an overall framework diagram of the method described in the present invention. DETAILED DESCRIPTION

[0060] In order to provide a further understanding and appreciation of the structural features and effects achieved by the present invention, a detailed description is provided with reference to preferred embodiments and accompanying drawings as follows:

[0061] like Figure 1 As shown, the present invention proposes a method for power transformer condition assessment based on a multi-scale gated attention fully convolutional network. This method proposes a transformer condition assessment method based on a multivariate time series classification model. The model consists of two parts: a multi-scale gated attention module and a fully convolutional network module. The designed multi-scale gated attention module first performs multi-scale convolution operations channel by channel through depthwise separable convolution to capture the multi-scale temporal information of different variables. Then, a gating layer is used to filter out the effective temporal patterns of different variables. Furthermore, the multi-scale features are fused with the original features, and the global dependencies of the time series are captured through multi-head self-attention. The multi-scale gated attention module can simultaneously capture the effective temporal information and global dependencies of different variables. Secondly, inspired by the anti-bottleneck structure, a fully convolutional network module is designed. This module first maps the original features to a high-dimensional space through a convolution operation, enriching the feature space and enhancing feature representation capabilities. Then, multi-layer convolution operations are used to increase the receptive field and capture cross-channel interactive dependencies between multiple variables.

[0062] The multi-scale gated attention module can capture the temporal patterns of different variables separately and retain effective temporal information through a gating mechanism. It also captures the global temporal dependencies of the sequence through multi-head self-attention. The fully convolutional network module can effectively learn the cross-channel interactive dependencies between multiple variables.

[0063] The specific steps are as follows:

[0064] The first step is to acquire and preprocess the multi-source sensor data of the power transformer: acquire the multi-source sensor data of the power transformer, including dissolved gas in oil, top oil temperature, core grounding current and power supply load data, and preprocess it.

[0065] The second step is the construction of the power transformer condition evaluation dataset: Based on the preprocessed multi-source sensor data, a two-dimensional matrix dataset of multiple variables and labels with time series characteristics is constructed as the power transformer condition evaluation dataset.

[0066] (1) The multi-source sensor data of the transformer, namely, dissolved gas in oil, trace water, core grounding current, top oil temperature, and power supply load data, will be obtained from the data monitoring platform of the substation. The data will be merged according to time to construct a complete data table.

[0067] (2) Preprocess the merged data table, manually analyze multiple variables separately, deal with missing values ​​and outlier problems of each variable, and then use the min-max normalization method to normalize the data to form a preprocessed data table.

[0068] (3) The preprocessed data table is integrated with the transformer operating status data according to the time dimension to construct a power transformer status evaluation dataset for training, where the last column is the label column and the remaining columns are feature columns.

[0069] The present invention uses multivariate time series data generated by multiple sensors to evaluate the health status of transformers. Therefore, the substation equipment status evaluation problem based on these multivariate time series data is defined as a classification problem based on multivariate time series. The multivariate time series classification problem is defined as follows: For a given multivariate time series dataset X with n samples = {X1, X2, ..., X n}, and label set Y={Y1,Y2,…,Y n}, where a single sample s represents the sequence length, c represents the number of variables; Y i ∈{1,2,…,m}m represents the number of categories. The goal of the multivariate time series classification task is to learn a multivariate time series X from the input i To category Y i The mapping relationship classifier.

[0070] The third step is to construct a power transformer state assessment model: a power transformer state assessment model is constructed based on a multi-scale gated attention module and a fully convolutional network module.

[0071] In order to effectively evaluate the status of substation equipment based on multivariate time series data, a multivariate time series classification network that integrates multi-scale gated attention and a fully convolutional network is proposed. This method can efficiently capture and utilize the temporal dependencies of multivariate time series data and the complex correlations and interactive dependencies between multiple variables. Figure 2As shown in the figure, the power transformer condition assessment model consists of two modules: a multi-scale gated attention (MSGA) and a fully convolutional network. Specifically, the MSGA extracts temporal patterns at different time scales through multi-scale convolution. Then, the MSGA passes through a gating unit, which dynamically adjusts the importance of features at different time scales based on contextual information. Finally, the multi-scale features are fused with the original features through multi-head self-attention to capture the global dependencies of the time series. Secondly, a fully convolutional network is used to capture the interactive dependencies between multiple variables. Finally, the features extracted by the MSGA and the FCN are fused to enrich the feature space and improve the model's classification performance.

[0072] (1) Construct a power transformer condition assessment model, which includes an input layer, a feature extraction layer, a feature fusion layer, and a classification layer.

[0073] (2) Set the input layer, which is used to read the original data and reconstruct the data into the data format used by the model.

[0074] (3) Set the feature extraction layer, which consists of a multi-scale gated attention module and a fully convolutional network module.

[0075] For multivariate time series data, different variables may exhibit distinct temporal patterns. Single-scale feature extraction may not fully capture the multivariate time series information, while redundant scale information can lead to reduced model performance. Therefore, a multi-scale gating mechanism is employed to extract multi-scale features of the sequence. This gating mechanism adaptively selects important scale information and removes redundant time scales. Furthermore, to effectively capture global temporal dependencies, multi-head self-attention is introduced. By fusing these two components, a multi-scale gated attention module is constructed. This module captures global temporal information and multi-scale information separately, and the fusion of these information enhances the model's multi-scale modeling capabilities for multivariate time series.

[0076] In order to make up for the deficiency of single-scale modeling in fully exploring the temporal regularity of each variable, we designed a multi-scale convolution module to extract the multi-scale information of each variable separately. We convolve the input features using a series of one-dimensional convolutional layers of varying sizes to capture local and global features at different scales. These multi-scale features are then concatenated and fused. Furthermore, to remove redundant timescale information and preserve the valid temporal patterns of each variable, we design a gating layer. This layer, applied after feature fusion, controls the information flow of the fused features and dynamically adjusts the importance of each time step and feature channel. The introduction of the gating mechanism gives the model greater flexibility and helps capture more complex time series patterns.

[0077] A1) Set up a multi-scale gated attention module, which consists of multi-scale convolution, gating mechanism and multi-head self-attention mechanism.

[0078] Multi-scale convolution is set to extract multi-scale temporal information of different variables. It consists of three one-dimensional convolutions with different convolution kernel sizes. The sizes of the three one-dimensional convolution kernels are set to 1*3, 1*5, and 1*7 respectively. The features extracted by the three one-dimensional convolutions are then spliced ​​to obtain the output of the multi-scale convolution module.

[0079]

[0080] in, Represents the multi-scale features after splicing and fusion, is the transpose of the input data, and DWC1d is a one-dimensional depth-wise separable convolution.

[0081] Depthwise Separable Convolution (DWConv) is an optimized convolution operation that can perform convolution operations on a channel-by-channel basis. To this end, we implemented a one-dimensional depthwise separable convolution (DWConv1d) and designed a multi-scale convolution module based on DWConv1d. In DWConv1d, each input time series variable is convolved with a separate convolution kernel to extract the time scale of a single variable, without any interaction between different variables. For the input feature map X, the convolution kernel shape is (C, K), where C is the number of input variables and K is the size of the convolution kernel.

[0082] Set the gating mechanism, which consists of a learnable one-dimensional vector, a sigmoid function, and a bitwise multiplication operation, as shown below:

[0083] F gate =σ(Conv1d(F ms ))⊙F ms ,

[0084] in, represents the gated weighted output feature, σ represents the Sigmoid activation function, and ⊙ is the element-wise multiplication operation;

[0085] Set up a multi-head self-attention mechanism, which consists of feature splicing operations and multi-head attention operations.

[0086]

[0087] in, represents the input features of multi-head self-attention, is the linear transformation matrix, are the queries, keys, and values ​​required for self-attention calculation after linear transformation. is a matrix multiplication operation, and d is the variable embedding dimension.

[0088] Long-term features reflect overall trends, while short-term features represent subtle changes in local regions, both of which are crucial. The multi-scale gating module can extract local temporal information from time series, but it cannot capture global dependencies. Therefore, we designed a multi-head self-attention algorithm based on multi-scale features to compensate for the lack of global dependency information.

[0089] A2) Set up a fully convolutional network module, which consists of three layers of convolution. The first and third layers of convolution are ordinary one-dimensional convolutions, and the second layer is a one-dimensional depth-wise separable convolution.

[0090] X′ i =Conv1d(DWC1d(Conv1d(X i )))

[0091] f=LN((X′ i ⊙X i )+X i ),

[0092] Among them, X i is the input feature, X′ i It is the output feature after three layers of convolution, and LN represents layer normalization.

[0093] To effectively capture the interactions and dependencies between multiple variables, a fully convolutional network was designed. Traditional convolutional architectures compress and then expand features when processing them. This operation can lead to the loss of important information during compression, resulting in a lack of useful information during feature extraction. Therefore, a fully convolutional feature extraction module was designed to overcome these shortcomings and effectively capture the dependencies between multiple variables.

[0094] (4) Setting the feature fusion layer,

[0095] The feature fusion layer consists of vector concatenation and vector addition operations. The output of the gating mechanism is concatenated with the output of the multi-head self-attention mechanism. A feedforward neural network is set to reconstruct the concatenated features. The reconstructed output features are then added to the output of the fully convolutional network module to achieve feature fusion.

[0096] (5) Set the classification layer and input the fused features into the MLP network for classification to obtain the status category of the power transformer, thereby realizing the status evaluation of the power transformer.

[0097] The fourth step is training the power transformer condition assessment model: the preprocessed power transformer basic data is sent to the power transformer condition assessment model for training.

[0098] (1) The constructed power transformer condition evaluation dataset is divided into a training set and a test set, and then the training set and the test set are reconstructed into data formats that can be read by the model, and the training set data is input into the input layer of the power transformer condition evaluation model.

[0099] (2) The feature extraction layer extracts effective features from the original data. The multi-scale gated attention module and the fully convolutional network module are used to extract the temporal features of the data and the dependencies between variables.

[0100] B1) Multi-scale gated attention module extracts temporal features:

[0101] Multi-scale convolution is used to extract multi-scale temporal information of multiple variables, and multi-scale features are obtained through feature concatenation. The multi-scale features are then input into a gating mechanism, which learns a gating vector. By multiplying the gating vector with the multi-scale features, the important temporal features of each variable are retained and redundant time scale information is filtered out.

[0102] The global temporal features of the original data are extracted through the multi-head self-attention mechanism, and the multi-scale features are fused with the original data features and input into the multi-head self-attention mechanism to learn and capture the global temporal features of the time series;

[0103] B2) Fully convolutional network module extracts dependencies between variables:

[0104] The original data is input into the fully convolutional network to extract the interactive dependencies between multiple variables.

[0105] (3) The feature fusion layer fuses multiple extracted effective features to obtain omnidirectional features:

[0106] The output features of the gating mechanism are fused with the output features of the multi-head self-attention mechanism to obtain the temporal features, and the temporal features are fused with the features output by the full convolutional network module to obtain the omnidirectional features.

[0107] (4) The classification layer performs classification prediction:

[0108] The extracted all-round features are used for classification prediction. The predicted value and the true value are calculated through the cross-entropy loss function to calculate the loss value. Then, the model parameters are updated through Adam and back propagation to continuously reduce the loss value until it stabilizes. The power transformer condition evaluation model is obtained. Subsequently, the model performance is evaluated using the test set. If the model performance meets the preset requirements, it is applied to the power transformer condition evaluation. Otherwise, the model hyperparameters need to be adjusted according to the evaluation results, and the model is retrained using the training set data until a power transformer condition evaluation model that meets the performance requirements is obtained.

[0109] When the time series data from multi-source sensors are input into the power transformer status evaluation model, the model will evaluate the current transformer status based on the data characteristics.

[0110] Step 5: Real-time monitoring of power transformers: Monitor power transformers in real time to obtain monitoring data and perform preprocessing.

[0111] Step 6: Evaluation of the power transformer status: Input the preprocessed real-time monitoring data into the trained power transformer status evaluation model to obtain the evaluation results of the power transformer status.

[0112] To verify the effectiveness of the method described in this paper, a comparative experiment was conducted with various methods on a test set. Experiments were conducted with sequence lengths of 24 and 48, and the experimental accuracy and F1 value indicators were recorded as shown in Table 1.

[0113] Table 1 Comparison of the method of the present invention with other methods

[0114]

[0115]

[0116] Through comparative experiments, it can be seen that the method of the present invention is significantly better than other methods in both experimental indicators. That is to say, the method of the present invention is more accurate and reliable than other methods in power transformer status evaluation.

[0117] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A power transformer condition assessment method based on a multi-scale gated attention fully convolutional network, characterized in that: The following steps are involved: 11) Acquisition and preprocessing of multi-source sensor data for power transformers: Acquisition and preprocessing of multi-source sensor data for power transformers, including dissolved gas in oil, top oil temperature, core grounding current, and power supply load data; 12) Construction of a power transformer condition evaluation dataset: Based on the preprocessed multi-source sensor data, a two-dimensional matrix dataset with multiple variables and labels with time series features is constructed as the power transformer condition evaluation dataset; 13) Construction of a power transformer condition assessment model: A power transformer condition assessment model is constructed based on a multi-scale gated attention module and a fully convolutional network module; The construction of the power transformer condition assessment model includes the following steps: 131) Constructing a power transformer condition assessment model, which includes an input layer, a feature extraction layer, a feature fusion layer, and a classification layer; 132) Set the input layer, which is used to read the original data and reconstruct the data into the data format used by the model; 133) Set the feature extraction layer, which consists of a multi-scale gated attention module and a fully convolutional network module. 1331) Set up a multi-scale gated attention module, which consists of multi-scale convolution, gating mechanism and multi-head self-attention mechanism, Multi-scale convolution is set to extract multi-scale temporal information of different variables. It consists of three one-dimensional convolutions with different convolution kernel sizes. The sizes of the three one-dimensional convolution kernels are set to 1*3, 1*5, and 1*7 respectively. The features extracted by the three one-dimensional convolutions are then spliced ​​to obtain the output of the multi-scale convolution module. in, Represents the multi-scale features after splicing and fusion, is the transpose of the input data, DWC1d is a one-dimensional depth-wise separable convolution; Set the gating mechanism, which consists of a learnable one-dimensional vector, a sigmoid function, and a bitwise multiplication operation, as shown below: F gate =σ(Conv1d(F ms ))⊙F ms , in, represents the gated weighted output feature, σ represents the Sigmoid activation function, and ⊙ is the element-wise multiplication operation; Set up a multi-head self-attention mechanism, which consists of feature splicing operations and multi-head attention operations. in, represents the input features of multi-head self-attention, is the linear transformation matrix, are the queries, keys, and values ​​required for self-attention calculation after linear transformation. is a matrix multiplication operation, and d is the variable embedding dimension; 1332) Set up a fully convolutional network module, which consists of three layers of convolution. The first and third layers of convolution are ordinary one-dimensional convolutions, and the second layer is a one-dimensional depth-separable convolution. X′ i =Conv1d(DWC1d(Conv1d(X i ))) f=LN((X′ i ⊙X i )+X i ), Among them, X i is the input feature, X i ′ is the output feature after three layers of convolution, and LN represents layer normalization; 134) Set the feature fusion layer, The feature fusion layer consists of vector concatenation and vector addition operations. It concatenates the output of the gating mechanism with the output of the multi-head self-attention mechanism. A feedforward neural network is set to reconstruct the concatenated features. The reconstructed output features are then added to the output of the fully convolutional network module to achieve feature fusion. 135) Setting a classification layer, inputting the fused features into the MLP network for classification, obtaining the status category of the power transformer, thereby realizing the status evaluation of the power transformer; 14) Training of the power transformer condition assessment model: The pre-processed power transformer basic data is fed into the power transformer condition assessment model for training; 15) Real-time monitoring of power transformers: Real-time monitoring of power transformers to obtain monitoring data and pre-process it; 16) Evaluation of power transformer status: The pre-processed real-time monitoring data is input into the trained power transformer status evaluation model to obtain the evaluation result of the power transformer status.

2. The power transformer state assessment method based on a multi-scale gated attention fully convolutional network according to claim 1 is characterized in that: The construction of the power transformer condition evaluation data set includes the following steps: 21) Obtain multi-source sensor data from the transformer from the substation's data monitoring platform, including dissolved gas in oil, trace water, core grounding current, top oil temperature, and power supply load data. Combine the data by time to construct a complete data table. 22) Preprocess the merged data table, manually analyze multiple variables separately, handle missing values ​​and outliers for each variable, and then use the min-max normalization method to normalize the data to form a preprocessed data table; 23) The preprocessed data table is integrated with the transformer operating status data according to the time dimension to construct a power transformer status evaluation dataset for training, where the last column is the label column and the remaining columns are feature columns.

3. The power transformer state assessment method based on a multi-scale gated attention fully convolutional network according to claim 1 is characterized in that: The training of the power transformer condition assessment model comprises the following steps: 31) dividing the constructed power transformer condition assessment data set into a training set and a test set, then reconstructing the training set and the test set into data formats readable by the model, and inputting the training set data into the input layer of the power transformer condition assessment model; 32) The feature extraction layer extracts effective features from the original data. The multi-scale gated attention module and the fully convolutional network module are used to extract the temporal features of the data and the dependencies between variables. 321) Multi-scale gated attention module extracts temporal features: Multi-scale convolution is used to extract multi-scale temporal information of multiple variables, and multi-scale features are obtained through feature concatenation. The multi-scale features are then input into a gating mechanism, which learns a gating vector. By multiplying the gating vector with the multi-scale features, the important temporal features of each variable are retained and redundant time scale information is filtered out. The global temporal features of the original data are extracted through the multi-head self-attention mechanism, and the multi-scale features are fused with the original data features and input into the multi-head self-attention mechanism to learn and capture the global temporal features of the time series; 322) The fully convolutional network module extracts dependencies between variables: The original data is input into the fully convolutional network to extract the interactive dependencies between multiple variables; 33) The feature fusion layer fuses the extracted multiple valid features to obtain omnidirectional features: The output features of the gating mechanism are fused with the output features of the multi-head self-attention mechanism to obtain the time features, and the time features are fused with the features output by the full convolutional network module to obtain the omnidirectional features. 34) The classification layer performs classification prediction: The extracted all-round features are used for classification prediction. The predicted value and the true value are calculated through the cross-entropy loss function. The model parameters are then updated through Adam and back-propagation to continuously reduce the loss value until it stabilizes. This results in a power transformer condition evaluation model. Subsequently, the model performance is evaluated using a test set. If the model performance meets the preset requirements, it is applied to power transformer condition evaluation. Otherwise, the model hyperparameters are adjusted based on the evaluation results, and the model is retrained using the training set data until a power transformer condition evaluation model that meets the performance requirements is obtained. When the time series data from multi-source sensors are input into the power transformer state evaluation model, the model evaluates the current transformer state based on the data characteristics.

Images