Method and System for Evaluating the Error State of Current Transformers Based on Multimodal Features

Abstract

The present invention belongs to the technical field of power monitoring, and discloses a method and system for evaluating the error state of a current transformer based on multi-modal features, including constructing a training data set; converting the training data into image data in multiple modalities, extracting global and local features from the image data in multiple modalities, and splicing the global and local features to obtain spliced features; splicing the global features and local features in multiple modalities to obtain full-modal global features and local features, calculating in combination with the original image data to obtain an attention feature map, operating on the attention feature map and the spliced features to obtain final fusion features; using the final fusion features for error state evaluation and outputting the error state situation. Based on the current time series data, the present invention efficiently fuses and utilizes the feature information of three different modalities, improving the reliability and accuracy of the evaluation results.

Description

Technical Field

[0001] The present invention relates to the technical field of power monitoring, and particularly to a method and system for evaluating the error state of a current transformer based on multi-modal features. Background Technique

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] In a power system, as a key device in the electric energy metering system, the accuracy of the metering performance of a transformer directly affects the fairness and reliability of electric energy transactions. A current transformer (CT), a precision device based on the principle of electromagnetic induction, can convert a high-magnitude primary current into a secondary current that is easy to measure. Its main structure includes a closed iron core and coils wound thereon. However, in the actual operating environment, the metering error of the CT may gradually deviate from the standard range over time due to the limitations of the acquisition mechanism and the influence of environmental conditions. This not only requires us to have accurate and rapid fault diagnosis capabilities, but also to be able to master the real-time state of the CT metering error at all times, so as to guide the maintenance personnel to reasonably plan the maintenance strategy. If the abnormality of the transformer state fails to be detected in time, it will surely pose a potential threat to the overall operation of the power grid.

[0004] Therefore, ensuring the real-time accuracy of current transformers, reducing electric energy metering losses, and ensuring the stable operation of measurement and control and protection equipment have become technical challenges that need to be solved urgently. How to achieve real-time online evaluation of the CT error state undoubtedly constitutes a key problem that the power industry urgently needs to overcome. The existing real-time online evaluation methods for CT error states only utilize the single-time series data characteristics of current transformer data and do not utilize the multi-modal data characteristic information. The evaluation results are not accurate and reliable enough. Summary of the Invention

[0005] To solve the above problems, the present invention proposes a method and system for evaluating the error state of a current transformer based on multi-modal features. On the basis of current time series data, three different modal feature information is efficiently fused and utilized, improving the reliability and accuracy of the evaluation results.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] In the first aspect, the present invention provides a method for evaluating the error state of a current transformer based on multi-modal features, including the following steps:

[0008] Collect the current data during the historical operation of normal current transformers and abnormal current transformers to construct a training data set;

[0009] Construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, extract global and local features from the image data in multiple modalities respectively, and splice the global and local features to obtain spliced features;

[0010] Gradually splice the global features and local features in multiple modalities respectively to obtain full-modal global features and full-modal local features. Calculate the attention based on the full-modal global features, full-modal local features and the original image data to obtain an attention feature map. After converting the attention feature map into a weight matrix, calculate it with the spliced features to adjust the importance degree of the feature points in the spliced features, and fuse the spliced features in multiple modalities after adjustment to obtain the final fused features;

[0011] Use the final fused features to evaluate the error state and output the error state of the current transformer.

[0012] As an alternative implementation, the multi-modal global and local fusion network includes a data modality conversion layer, a global and local feature extraction layer, a multi-modal global and local feature fusion layer and a fully connected layer.

[0013] As an alternative implementation, input the training data into the multi-modal global and local fusion network and convert it into image data in multiple modalities. Specifically:

[0014] Convert the training data into two-dimensional image data in three modalities through three algorithms: GAF, MTF and RPM. GAF converts the training data into a matrix representation in the angular domain, MTF converts the training data into a probability matrix through a Markov process, and RPM generates a reproduction map by recording the distance relationship between the sequence points of the training data.

[0015] As an alternative implementation, extract global features through global average pooling, and then use pointwise convolution to adjust its channels;

[0016] Extract local features through pointwise convolution, and at the same time use a residual structure to connect the original image data of the same size to maintain the original context information.

[0017] As an alternative implementation, after the attention feature map is converted into a feature weight matrix through the sigmoid function, perform a Hadamard product with the spliced features corresponding to the three modalities to adjust the importance degree of the feature points in each spliced feature.

[0018] As an alternative implementation, use the final fused features to evaluate the error state. Specifically:

[0019] The final fused features are fed into a fully connected layer, integrated into the corresponding probability distribution, and classified using an SVM classifier to output the error status of the current current transformer.

[0020] In a second aspect, the present invention provides a current transformer error status evaluation system based on multi-modal features, including:

[0021] A training set construction module, configured to: collect the current data of normal and abnormal current transformers during historical operation, and construct a training data set;

[0022] A feature extraction module, configured to: construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, respectively extract global and local features from the image data in multiple modalities, and splice the global and local features to obtain spliced features;

[0023] A feature fusion module, configured to: respectively splice the global features and local features in multiple modalities step by step to obtain full-modal global features and full-modal local features, calculate attention based on the full-modal global features, full-modal local features, and the original image data to obtain an attention feature map, convert the attention feature map into a weight matrix and calculate it with the spliced features, adjust the importance of feature points in the spliced features, and fuse the adjusted spliced features in multiple modalities to obtain the final fused features;

[0024] An error status evaluation module, configured to: use the final fused features to evaluate the error status and output the error status of the current transformer.

[0025] In a third aspect, the present invention provides an electronic device, including a memory and a processor, as well as computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the following steps are completed:

[0026] Collect the current data of normal and abnormal current transformers during historical operation, and construct a training data set;

[0027] Construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, respectively extract global and local features from the image data in multiple modalities, and splice the global and local features to obtain spliced features;

[0028] The global features and local features in multiple modalities are respectively concatenated level by level to obtain full-modal global features and full-modal local features. Attention calculation is performed based on the full-modal global features, full-modal local features, and the original image data to obtain an attention feature map. After converting the attention feature map into a weight matrix, it is calculated with the concatenated features to adjust the importance of the feature points in the concatenated features. The concatenated features in multiple modalities after adjustment are fused to obtain the final fused features;

[0029] The error state of the current transformer is evaluated using the final fused features, and the error state condition of the current transformer is output.

[0030] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the following steps are completed:

[0031] Collect the current data during the historical operation of normal current transformers and abnormal current transformers to construct a training data set;

[0032] Construct a multi-modal global and local fusion network. Input the training data into the multi-modal global and local fusion network to convert it into image data in multiple modalities. Respectively extract the global and local features of the image data in multiple modalities, and concatenate the global and local features to obtain concatenated features;

[0033] The global features and local features in multiple modalities are respectively concatenated level by level to obtain full-modal global features and full-modal local features. Attention calculation is performed based on the full-modal global features, full-modal local features, and the original image data to obtain an attention feature map. After converting the attention feature map into a weight matrix, it is calculated with the concatenated features to adjust the importance of the feature points in the concatenated features. The concatenated features in multiple modalities after adjustment are fused to obtain the final fused features;

[0034] The error state of the current transformer is evaluated using the final fused features, and the error state condition of the current transformer is output.

[0035] In a fifth aspect, the present invention provides a computer program product including a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0036] Collect the current data during the historical operation of normal current transformers and abnormal current transformers to construct a training data set;

[0037] Construct a multi-modal global and local fusion network. Input the training data into the multi-modal global and local fusion network to convert it into image data in multiple modalities. Respectively extract the global and local features of the image data in multiple modalities, and concatenate the global and local features to obtain concatenated features;

[0038] The global features and local features in multiple modalities are concatenated level by level respectively to obtain the full-modal global features and full-modal local features. Attention calculation is performed based on the full-modal global features, full-modal local features, and the original image data to obtain an attention feature map. After converting the attention feature map into a weight matrix, it is calculated with the concatenated features to adjust the importance of feature points in the concatenated features. The concatenated features in multiple modalities after adjustment are fused to obtain the final fused features;

[0039] The error state of the current transformer is evaluated using the final fused features, and the error state of the current transformer is output.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0041] The present disclosure proposes a method and system for evaluating the error state of a current transformer based on multi-modal features, designs a multi-modal global and local feature fusion method, converts one-dimensional current time-series data into two-dimensional image features in three modalities, extracts and fuses the image features in different modalities respectively to obtain a comprehensive feature that fuses different modality information, and uses the comprehensive feature to evaluate the error of the current transformer. Based on the current time-series data, the present invention efficiently fuses and utilizes the feature information of three different modalities, comprehensively analyzes and utilizes the internal structure and dynamic characteristics of the current time-series data, thereby improving the reliability and accuracy of the error evaluation result.

[0042] The present disclosure proposes a method and system for evaluating the error state of a current transformer based on multi-modal features, and proposes a multi-modal global and local feature fusion network structure. The network mainly consists of a data modality conversion layer, a global and local feature extraction layer, a multi-modal global and local feature fusion layer, and a fully connected layer. The data modality conversion layer converts one-dimensional current time-series data into two-dimensional image data in three modalities; the global and local feature extraction layer extracts global features and local features from the multi-modal two-dimensional images; the multi-modal global and local feature fusion layer efficiently fuses the concatenated features of features in different modalities. Finally, the fused comprehensive feature is sent into the fully connected layer to integrate the comprehensive feature into the corresponding probability distribution, and an SVM classifier is used for classification to output the error state of the current transformer, realizing real-time online evaluation of the error state of the current transformer.

[0043] Advantages of additional aspects of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings constituting a part of this specification are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0045] Figure 1 It is a flowchart of the method for evaluating the error state of a current transformer based on multi-modal features provided in Embodiment 1 of the present invention;

[0046] Figure 2 It is a confusion matrix of the result of identifying the error state of a current transformer by using the method of the present invention. Detailed implementation manners

[0047] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0048] It should be noted that the following detailed description is exemplary and is intended to provide further illustration of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0049] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units need not be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0050] In the case of no conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other.

[0051] Term explanation:

[0052] Gramian Angular Field (GAF): It is an effective method for converting one-dimensional time series data into a two-dimensional image representation. Its core principle is to calculate the cosine values of the angles between data points in the time series and map these values to the pixels of a two-dimensional image, thereby generating an image that can reflect the dynamic and periodic characteristics of the time series.

[0053] Markov Transition Field (MTF): It is a method for converting time series data into two-dimensional images, mainly used in the fields of computer vision and image processing, especially in video sequence analysis. MTF is based on the concept of Markov chains. By modeling the probability transitions between states in a time series, it converts time series data into two-dimensional images, thereby providing local information about the signal in both time and frequency. The basic principle is that in MTF, a video sequence is divided into a grid-like cell, and the transitions between pixel intensity values in each cell are modeled as a Markov chain. By calculating and storing the transition probabilities between each pixel intensity value, these probabilities are stored in a transition matrix, which can be visualized through a heatmap, showing the transition probabilities from one intensity value to another. This heatmap is called the Markov transition field, providing a compact representation of the spatio-temporal patterns in the video sequence.

[0054] Relative Position Matrix (RPM): It is a matrix representation method used to describe the relative position relationships between objects. It converts the relative position relationships between objects into data through a mathematical model, providing a basis for subsequent analysis and applications. The relative position matrix is established based on the coordinate system of a two-dimensional plane or three-dimensional space. In a determined coordinate system, the position of each object can be represented by coordinates, and the relative position is obtained by calculating the coordinate differences between different objects. The relative position matrix presents these differences in matrix form. For example, in a two-dimensional plane, if there are n objects, the position of each object can be represented as a two-dimensional coordinate (x, y), then the size of the relative position matrix is n×n, and the element in the i-th row and j-th column represents the position difference of the i-th object relative to the j-th object.

[0055] Embodiment 1

[0056] As Figure 1 shown, this embodiment provides a method for evaluating the error state of a current transformer based on multi-modal features, including the following steps:

[0057] S1. Collect the current data during the historical operation of normal and abnormal current transformers to construct a training data set;

[0058] S2. Construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, extract global and local features from the image data in multiple modalities respectively, and splice the global and local features to obtain spliced features;

[0059] S3, respectively, splicing the global features and local features under multiple modalities step by step to obtain full-modal global features and full-modal local features, performing attention calculation based on the full-modal global features, full-modal local features and original image data to obtain an attention feature map, converting the attention feature map into a weight matrix and then calculating it with the splicing features, adjusting the importance of feature points in the splicing features, and fusing the adjusted splicing features under multiple modalities to obtain the final fusion feature;

[0060] S4. Use the final fusion features to evaluate the error state and output the error state of the current transformer.

[0061] The general idea of ​​the present invention is: first, collect the secondary current data of multiple current transformers during historical operation, including normal current transformers and abnormal current transformers, and use the historical current data samples (one-dimensional current time series data) provided by normal and abnormal current transformers to build a training data set. Then, construct a multimodal global and local fusion network, which is mainly composed of a data mode conversion layer, a global and local feature extraction layer, a multimodal global and local feature fusion layer, and a fully connected layer. In the data mode conversion layer, the one-dimensional current time series data will be converted into two-dimensional image data of three modes through three algorithms: Gramian Angular Field (GAF), Markov Transition Field (MTF), and Relative Position Matrix (RPM); in the global and local feature extraction layer, the two-dimensional image data under the three modes will be extracted globally and locally; the three sets of global and local multimodal features extracted will be globally and locally efficiently fused with different modal features in the multimodal global and local feature fusion layer. Finally, the fused features are sent to the fully connected layer to accurately evaluate the error state of the current transformer. In practical applications, the network is first trained with a large number of data samples, and then the trained network is connected to the secondary current data of the current transformer to be tested, so that the error state of the current transformer can be evaluated online in real time.

[0062] (1) Data modality conversion layer

[0063] After the one-dimensional current time-series data is input into the multi-modal global and local fusion network, it will first undergo data modal conversion through the data modal conversion layer. The data modal conversion layer mainly converts the one-dimensional data into two-dimensional image data in three modalities through three algorithms: GAF, MTF, and RPM. GAF mainly converts the time-series data into a matrix representation in the angular domain, capturing the periodicity and phase information of the data; MTF converts the sequence into a probability matrix through a Markov process, reflecting the state transition characteristics of the sequence; RPM generates a recurrence plot by recording the distance relationship between sequence points, revealing the complexity and self-similarity of the sequence. The process can be expressed as:

[0064]

[0065]

[0066] In the formula, X is the one-dimensional current time-series data input into the network, , and are the two-dimensional image data obtained through the conversion of the GAF, MTF, and RPM algorithms, respectively.

[0067] (2) Global and local feature extraction layer

[0068] In the global and local feature extraction layer, the three groups of two-dimensional image data after modal conversion , and are used to extract global features and local features respectively and then concatenate and output. In the global feature extraction part, first, global features are extracted through global average pooling, and then pointwise convolution is used to adjust its channels. Taking as an example for global feature extraction, its calculation process can be expressed as follows:

[0069]

[0070] In the formula, represents the global feature extracted from , is the two-dimensional image feature obtained after modal transformation by the GAF algorithm, represents the global pooling operation, represents the pointwise convolution operation. After global feature extraction is performed on and , the corresponding global features and can also be obtained.

[0071] In the local feature extraction part, local features are mainly extracted through pointwise convolution, and at the same time, the residual structure is used to connect the original features of the same size to maintain the original context information. Taking as an example of local feature extraction, its calculation process can be expressed as follows:

[0072]

[0073] In the formula, is the two-dimensional image feature obtained after the modal transformation by the GAF algorithm, represents the local feature extracted from , represents the pointwise convolution operation. After local feature extraction is performed on and , the corresponding global features and can also be obtained respectively. Subsequently, the obtained global features and the corresponding local features are concatenated to obtain three sets of concatenated multi-modal global and local features (i.e., concatenated features). The process can be expressed as:

[0074]

[0075]

[0076]

[0077] In the formula, is the concatenated feature extracted after passes through the global and local feature extraction layers, is the concatenated feature extracted after passes through the global and local feature extraction layers, is the concatenated feature extracted after passes through the global and local feature extraction layers, Concat(;) is the feature concatenation operation.

[0078] (3) Multi-modal global and local feature fusion layer

[0079] In the multi-modal global and local feature fusion layer, the global feature parts of the three modalities are concatenated step by step to obtain the global feature containing all-modal information (i.e., the all-modal global feature). The local feature parts of the three modalities are concatenated step by step to obtain the local feature containing all-modal information (i.e., the all-modal local feature). Subsequently, through the initial information (original image data) combined with and Calculate the attention mechanism and output the attention feature maps generated from the initial information (i.e., the original image data) in three modalities and the corresponding global and local information. Through this attention feature map, after sigmoid the function is converted into a feature weight matrix in the range (0,1), and then the Hadamard product is performed with the global and local features (i.e., the concatenated features) corresponding to the three modalities to adjust the importance of the feature points in each global and local feature (i.e., the concatenated features). Finally, the adjusted global and local features (i.e., the concatenated features) in the three modalities are reshaped in scale and then superimposed and fused to output the final fused features . This process can be expressed by the following formula:

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086] In the above formula, , and are the global features in three different modalities respectively, is the global feature containing all-modal information after concatenating the global features in the three modalities. , and are the local features in three different modalities respectively, is the local feature containing all-modal information after concatenating the local features in the three modalities. , , are the initial information (i.e., the original image data) of the three modalities respectively , and The three feature weight matrices obtained by converting the attention feature maps through sigmoid the function. 、 、 are all dimension mapping functions, and their role is to convert the dimensions of the multi-modal features and the original information participating in the calculation to a unified dimension d for calculation. r is a scale reshaping function, and its role is to reshape the feature scales in different modalities to be consistent for easy superimposition and fusion. is the final output feature of the multi-modal global and local fusion network. ⊙ represents the Hadamard product operation. Softmax is the normalized exponential function.

[0087] (4)Fully connected layer

[0088] Subsequently, the fused features are fed into the fully connected layer to integrate the comprehensive features into the corresponding probability distribution, and the SVM classifier is used for classification to output the error state of the current current transformer.

[0089] The multi-modal global and local fusion network is offline trained with the current transformer data collected, and after obtaining a fully trained network, the secondary side current signal of the current transformer to be measured is connected to perform real-time error state evaluation to obtain the error state of the current transformer to be measured.

[0090] The following are the specific implementation effects of the method of the present invention, and experiments are carried out with the collected historical current data samples. For 300 groups of experimental samples, including 150 normal current transformer samples and 150 abnormal current transformer samples, the error state identification is carried out using the method of the present invention, and the confusion matrix of the obtained identification results is as Figure 2 shown:

[0091] The confusion matrix shows that for the 150 normal samples in the test samples, 149 normal samples are correctly evaluated by the algorithm, and for the 150 abnormal samples in the test samples, 150 abnormal samples are correctly evaluated by the algorithm. According to this result, it can be calculated that the accuracy rate of the algorithm reaches 99.67%, indicating that the method of the present invention can accurately identify the error state of the current transformer.

[0092] The present invention designs a multi-modal global and local feature fusion method, which converts one-dimensional current time series data into two-dimensional image features in three modalities, extracts and fuses the image features in different modalities respectively to obtain comprehensive features integrating different modality information, and uses the comprehensive features for error evaluation of the current transformer. Based on the current time series data, the present invention efficiently fuses and utilizes the feature information of three different modalities, comprehensively analyzes and utilizes the internal structure and dynamic characteristics of the current time series data, thereby improving the reliability and accuracy of the error evaluation result.

[0093] The present invention proposes a multi-modal global and local feature fusion network structure, which mainly consists of a data modality conversion layer, a global and local feature extraction layer, a multi-modal global and local feature fusion layer, and a fully connected layer. The data modality conversion layer converts one-dimensional current time series data into two-dimensional image data in three modalities; the global and local feature extraction layer extracts global features and local features from the multi-modal two-dimensional images; the multi-modal global and local feature fusion layer performs efficient fusion of the concatenated features of the features in different modalities. Finally, the fused comprehensive features are sent to the fully connected layer to integrate the comprehensive features into the corresponding probability distribution, and an SVM classifier is used for classification to output the error state of the current current transformer, realizing real-time online evaluation of the error state of the current transformer.

[0094] Embodiment 2

[0095] This embodiment provides a current transformer error state evaluation system based on multi-modal features, including:

[0096] A training set construction module, configured to: collect the current data during the historical operation of normal current transformers and abnormal current transformers, and construct a training data set;

[0097] A feature extraction module, configured to: construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, respectively extract global features and local features from the image data in multiple modalities, and concatenate the global and local features to obtain concatenated features;

[0098] A feature fusion module, configured to: respectively concatenate the global features and local features in multiple modalities step by step to obtain full-modal global features and full-modal local features, perform attention calculation based on the full-modal global features, full-modal local features, and the original image data to obtain an attention feature map, convert the attention feature map into a weight matrix and calculate it with the concatenated features, adjust the importance of the feature points in the concatenated features, and fuse the adjusted concatenated features in multiple modalities to obtain the final fused features;

[0099] An error state evaluation module, configured to: use the final fused features for error state evaluation and output the error state of the current transformer.

[0100] It should be noted here that the above modules correspond to the steps described in Embodiment 1. The examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the content disclosed in the above Embodiment 1. It should be noted that the above modules, as part of the system, can be executed in a computer system such as a set of computer executable instructions.

[0101] In more embodiments, there is also provided:

[0102] An electronic device includes a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the method described in Embodiment 1 is completed:

[0103] Collect the current data during the historical operation of the normal current transformer and the abnormal current transformer, and construct a training data set;

[0104] Construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, extract global and local features from the image data in multiple modalities respectively, and splice the global and local features to obtain spliced features;

[0105] Splice the global features and local features in multiple modalities step by step respectively to obtain full-modal global features and full-modal local features. Calculate the attention based on the full-modal global features, full-modal local features, and the original image data to obtain an attention feature map. After converting the attention feature map into a weight matrix, calculate it with the spliced features to adjust the importance of the feature points in the spliced features, and fuse the adjusted spliced features in multiple modalities to obtain the final fused features;

[0106] Use the final fused features to evaluate the error state and output the error state of the current transformer.

[0107] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0108] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0109] A computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the following method is completed:

[0110] Collect the current data during the historical operation of the normal current transformer and the abnormal current transformer, and construct a training data set;

[0111] Construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, respectively extract global and local features from the image data in multiple modalities, and splice the global and local features to obtain spliced features;

[0112] Respectively splice the global features and local features in multiple modalities step by step to obtain full-modal global features and full-modal local features, perform attention calculation based on the full-modal global features, full-modal local features and the original image data to obtain an attention feature map, convert the attention feature map into a weight matrix and calculate it with the spliced features to adjust the importance of the feature points in the spliced features, and fuse the adjusted spliced features in multiple modalities to obtain the final fusion features;

[0113] Use the final fusion features to evaluate the error state and output the error state of the current transformer.

[0114] The method in Embodiment 1 can be directly embodied as being executed by a hardware processor, or by a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0115] A computer program product includes a computer program, and when the computer program is executed by a processor, the following method is implemented:

[0116] Collect the current data during the historical operation of normal current transformers and abnormal current transformers to construct a training data set;

[0117] Construct a multi-modal global and local fusion network, input the training data into the multi-modal global and local fusion network, convert it into image data in multiple modalities, respectively extract global and local features from the image data in multiple modalities, and splice the global and local features to obtain spliced features;

[0118] Respectively splice the global features and local features in multiple modalities step by step to obtain full-modal global features and full-modal local features, perform attention calculation based on the full-modal global features, full-modal local features and the original image data to obtain an attention feature map, convert the attention feature map into a weight matrix and calculate it with the spliced features to adjust the importance of the feature points in the spliced features, and fuse the adjusted spliced features in multiple modalities to obtain the final fusion features;

[0119] The error state is evaluated using the final fused features, and the error state of the current transformer is output.

[0120] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which are executed in a device on a target real or virtual processor to perform the processes / methods described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules can be combined or divided as needed between program modules. The machine-executable instructions for program modules can be executed within a local or distributed device. In a distributed device, program modules can be located in local and remote storage media.

[0121] The computer program code for implementing the method of the present invention can be written in one or more programming languages. This computer program code can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program code is executed by the computer or other programmable data processing device, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the computer, partially on the computer, as a stand-alone software package, partially on the computer and partially on a remote computer, or entirely on a remote computer or server.

[0122] In the context of the present invention, the computer program code or related data can be carried by any suitable carrier so that a device, apparatus, or processor can perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals can include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, etc.

[0123] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with this embodiment can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0124] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, it is not a limitation on the protection scope of the present invention. Those skilled in the art should understand that, based on the technical solution of the present invention, various modifications or deformations that can be made by those skilled in the art without creative efforts are still within the protection scope of the present invention.

Claims

1. A current transformer error state evaluation method based on multi-modal characteristics, characterized in that: The following steps are involved: Collect the current data of normal current transformers and abnormal current transformers during historical operation to build a training data set; Construct a multimodal global and local fusion network, input the training data into the multimodal global and local fusion network, convert it into image data under multiple modalities, extract global and local features from the image data under multiple modalities respectively, and splice the global and local features to obtain spliced ​​features; The global features and local features under multiple modalities are spliced ​​step by step to obtain full-modal global features and full-modal local features. Attention calculation is performed based on the full-modal global features, full-modal local features and original image data to obtain an attention feature map. The attention feature map is converted into a weight matrix and then calculated with the splicing features. The importance of the feature points in the splicing features is adjusted, and the splicing features under multiple modalities after adjustment are fused to obtain the final fused features. The final fusion feature is used to evaluate the error state and output the error state of the current transformer; The multimodal global and local fusion network includes a data modality conversion layer, a global and local feature extraction layer, a multimodal global and local feature fusion layer, and a fully connected layer; The training data is input into the multimodal global and local fusion network and converted into image data under multiple modalities, specifically: The training data are converted into two-dimensional image data under three modes through three algorithms: GAF, MTF and RPM. GAF converts the training data into a matrix representation in the angular domain. MTF converts the training data into a probability matrix through a Markov process. RPM generates a reproduction graph by recording the distance relationship between the sequence points of the training data.

2. The current transformer error state evaluation method based on multi-modal features according to claim 1, characterized in that: The global features are extracted by global average pooling, and then the channels are adjusted by point-by-point convolution. Local features are extracted through point-by-point convolution, and the residual structure is used to connect the original image data of the same size to maintain the original context information.

3. The current transformer error state evaluation method based on multi-modal features according to claim 1, characterized in that: The attention feature map is converted into a feature weight matrix through the sigmoid function and then Hadamard product is performed with the concatenated features corresponding to the three modalities to adjust the importance of the feature points in each concatenated feature.

4. The current transformer error state evaluation method based on multi-modal features according to claim 1, characterized in that: The final fusion features are used to evaluate the error state, specifically: The final fusion features are sent to the fully connected layer, integrated into the corresponding probability distribution, and classified using the SVM classifier to output the error state of the current current transformer.

5. A current transformer error state evaluation system based on multi-modal characteristics, characterized in that: include: The training set building module is configured to: collect current data of normal current transformers and abnormal current transformers during historical operation to build a training data set; The feature extraction module is configured to: construct a multimodal global and local fusion network, input the training data into the multimodal global and local fusion network, convert it into image data under multiple modalities, extract global and local features from the image data under multiple modalities respectively, and splice the global and local features to obtain spliced ​​features; The feature fusion module is configured to: respectively splice the global features and local features under multiple modalities step by step to obtain full-modal global features and full-modal local features, perform attention calculation based on the full-modal global features, the full-modal local features and the original image data to obtain an attention feature map, convert the attention feature map into a weight matrix and then calculate it with the splicing feature, adjust the importance of feature points in the splicing feature, and fuse the adjusted splicing features under multiple modalities to obtain the final fusion feature; The error state evaluation module is configured to: use the final fusion feature to perform error state evaluation and output the error state of the current transformer; The multimodal global and local fusion network includes a data modality conversion layer, a global and local feature extraction layer, a multimodal global and local feature fusion layer, and a fully connected layer; The training data is input into the multimodal global and local fusion network and converted into image data under multiple modalities, specifically: The training data are converted into two-dimensional image data under three modes through three algorithms: GAF, MTF and RPM. GAF converts the training data into a matrix representation in the angular domain. MTF converts the training data into a probability matrix through a Markov process. RPM generates a reproduction graph by recording the distance relationship between the sequence points of the training data.

6. An electronic device, characterized in that: The invention comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein the computer instructions are executed by the processor to complete the following steps: Collect the current data of normal current transformers and abnormal current transformers during historical operation to build a training data set; Construct a multimodal global and local fusion network, input the training data into the multimodal global and local fusion network, convert it into image data under multiple modalities, extract global and local features from the image data under multiple modalities respectively, and splice the global and local features to obtain spliced ​​features; The global features and local features under multiple modalities are spliced ​​step by step to obtain full-modal global features and full-modal local features. Attention calculation is performed based on the full-modal global features, full-modal local features and original image data to obtain an attention feature map. The attention feature map is converted into a weight matrix and then calculated with the splicing features. The importance of the feature points in the splicing features is adjusted, and the splicing features under multiple modalities after adjustment are fused to obtain the final fused features. The final fusion feature is used to evaluate the error state and output the error state of the current transformer; The multimodal global and local fusion network includes a data modality conversion layer, a global and local feature extraction layer, a multimodal global and local feature fusion layer, and a fully connected layer; The training data is input into the multimodal global and local fusion network and converted into image data under multiple modalities, specifically: The training data are converted into two-dimensional image data under three modes through three algorithms: GAF, MTF and RPM. GAF converts the training data into a matrix representation in the angular domain. MTF converts the training data into a probability matrix through a Markov process. RPM generates a reproduction graph by recording the distance relationship between the sequence points of the training data.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the following steps are implemented: Collect the current data of normal current transformers and abnormal current transformers during historical operation to build a training data set; Construct a multimodal global and local fusion network, input the training data into the multimodal global and local fusion network, convert it into image data under multiple modalities, extract global and local features from the image data under multiple modalities respectively, and splice the global and local features to obtain spliced ​​features; The global features and local features under multiple modalities are spliced ​​step by step to obtain full-modal global features and full-modal local features. Attention calculation is performed based on the full-modal global features, full-modal local features and original image data to obtain an attention feature map. The attention feature map is converted into a weight matrix and then calculated with the splicing features. The importance of the feature points in the splicing features is adjusted, and the splicing features under multiple modalities after adjustment are fused to obtain the final fused features. The final fusion feature is used to evaluate the error state and output the error state of the current transformer; The multimodal global and local fusion network includes a data modality conversion layer, a global and local feature extraction layer, a multimodal global and local feature fusion layer, and a fully connected layer; The training data is input into the multimodal global and local fusion network and converted into image data under multiple modalities, specifically: The training data are converted into two-dimensional image data under three modes through three algorithms: GAF, MTF and RPM. GAF converts the training data into a matrix representation in the angular domain. MTF converts the training data into a probability matrix through a Markov process. RPM generates a reproduction graph by recording the distance relationship between the sequence points of the training data.

8. A computer program product, stored on a computer-readable storage medium, characterized in that: The invention comprises a computer program, which, when executed by a processor, implements the following steps: Collect the current data of normal current transformers and abnormal current transformers during historical operation to build a training data set; Construct a multimodal global and local fusion network, input the training data into the multimodal global and local fusion network, convert it into image data under multiple modalities, extract global and local features from the image data under multiple modalities respectively, and splice the global and local features to obtain spliced ​​features; The global features and local features under multiple modalities are spliced ​​step by step to obtain full-modal global features and full-modal local features. Attention calculation is performed based on the full-modal global features, full-modal local features and original image data to obtain an attention feature map. The attention feature map is converted into a weight matrix and then calculated with the splicing features. The importance of the feature points in the splicing features is adjusted, and the splicing features under multiple modalities after adjustment are fused to obtain the final fused features. The final fusion feature is used to evaluate the error state and output the error state of the current transformer; The multimodal global and local fusion network includes a data modality conversion layer, a global and local feature extraction layer, a multimodal global and local feature fusion layer, and a fully connected layer; The training data is input into the multimodal global and local fusion network and converted into image data under multiple modalities, specifically: The training data are converted into two-dimensional image data under three modes through three algorithms: GAF, MTF and RPM. GAF converts the training data into a matrix representation in the angular domain. MTF converts the training data into a probability matrix through a Markov process. RPM generates a reproduction graph by recording the distance relationship between the sequence points of the training data.

Images