Mutual inductor metering error online detection method and device

Abstract

The invention discloses a mutual inductor metering error online detection method and device, and relates to the technical field of power grid operation and maintenance, and the method comprises the following steps: collecting historical data, carrying out the preprocessing, and obtaining voltage historical data and phase historical data; preprocessing the voltage historical data and the phase historical data, and constructing a training set and a test set; constructing a metering error online detection model; training a metering error online detection model by using the training set and the test set; inputting to-be-predicted data into the trained metering error online detection model to obtain a corresponding prediction result; and performing metering error evaluation on the target mutual inductor based on the prediction result. The advantages of the deep learning technology in the aspects of feature extraction and mode recognition can be fully utilized, higher accuracy and robustness are achieved, and the method is suitable for mutual inductor metering error online detection tasks in various industrial and energy fields.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid operation and maintenance, and in particular to an online detection method and device for mutual inductor metering errors. Background Art

[0002] The efficient, stable, and reliable operation of power systems is crucial to the national economy and livelihood, and is the foundation of national economic development. As a key signal acquisition device in power systems, instrument transformers (CTs) ensure reliable electrical isolation between the primary high-voltage system and secondary equipment, accurately measuring primary voltage and current while ensuring the safety of secondary equipment and electricity consumption. This provides a reliable basis for energy metering, condition monitoring, and relay protection. Capacitor voltage transformers (CVTs) are widely used in power systems of 110 kV and above due to their excellent insulation performance and cost-effectiveness. However, compared to traditional electromagnetic voltage transformers, these voltage transformers are more complex in structure and have lower error stability, making them prone to out-of-tolerance errors during actual operation. Field operation experience shows that, among instrument transformers at voltage levels of 110 kV and above, the failure rate of CVTs (capacitor voltage transformers) is approximately five times that of PTs (voltage transformers) and ten times that of electromagnetic current transformers. As a measuring device, the long-term stability of measurement error is one of the most important parameters for evaluating CVT performance.

[0003] Due to the difficulty in preventing power outages on high-voltage transmission lines, a large number of CVTs in the power grid are operating beyond their inspection period, and there is a risk of excessive metering errors, which affects the fair trade settlement of electricity. Existing metering error state assessment methods are no longer suitable for the online monitoring of key equipment status in smart substations. Therefore, it is necessary to conduct research on methods for evaluating and predicting the metering error state of in-service CVTs without power outages, so as to grasp the metering error state of CVTs in real time and provide more targeted guidance on the operation and maintenance of CVTs. This is of great significance for ensuring the safe, stable and economical operation of the power system. At the same time, the relevant technical routes and research methods for evaluating and predicting the metering error state of in-service CVTs can be extended to the research of other types of power transformers, which has important reference value for promoting the development of industry technology.

[0004] Therefore, in order to meet actual needs, an online detection technology for mutual inductor measurement error is provided. Summary of the Invention

[0005] In response to the defects existing in the prior art, the purpose of the present invention is to provide a method and device for online detection of transformer measurement errors, which can fully utilize the advantages of deep learning technology in feature extraction and pattern recognition, have higher accuracy and robustness, and are suitable for online detection tasks of transformer measurement errors in various industrial and energy fields.

[0006] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0007] In a first aspect, the present application provides a method for online detection of mutual inductor measurement error, the method comprising the following steps:

[0008] Collect historical data, perform preprocessing, and obtain voltage historical data and phase historical data;

[0009] Preprocessing the voltage history data and the phase history data, and constructing a training set and a test set;

[0010] Construct an online detection model for measurement errors;

[0011] Using the training set and the test set, training the measurement error online detection model;

[0012] Inputting the data to be predicted into the trained online measurement error detection model to obtain the corresponding prediction result;

[0013] Based on the prediction results, the target mutual inductor is evaluated for measurement error; wherein,

[0014] The data to be predicted includes voltage data to be predicted and phase data to be predicted.

[0015] On the basis of the above technical solution, the voltage history data and the phase history data are preprocessed, and a training set and a test set are constructed, including the following steps:

[0016] Based on historical data, construct the original data feature table;

[0017] Normalizing the original data feature table to obtain a processed summary data table;

[0018] Based on the processed summary data table, the voltage history data and the phase history data are acquired.

[0019] On the basis of the above technical solution, the voltage history data and the phase history data are preprocessed, and a training set and a test set are constructed, including the following steps:

[0020] Missing value processing, outlier processing, duplicate data removal processing, and normalization processing are performed on the voltage history data and the phase history data to obtain preprocessed voltage history data and preprocessed phase history data.

[0021] On the basis of the above technical solution, the construction of the online measurement error detection model includes the following steps:

[0022] Based on Transformer and CNN, an online detection model for measurement errors is constructed.

[0023] On the basis of the above technical solution, the construction of the online measurement error detection model includes the following steps:

[0024] Based on the training set and the test set, extracting local features corresponding to the time series through CNN;

[0025] Based on local features, global information modeling is performed through Transformer to obtain an online detection model for measurement errors.

[0026] In a second aspect, the present application provides an online detection device for mutual inductor measurement error, the device comprising:

[0027] The data acquisition module is used to collect historical data, perform preprocessing, and obtain voltage historical data and phase historical data;

[0028] A data set construction module, which is used to preprocess the voltage history data and the phase history data and construct a training set and a test set;

[0029] A model building module, which is used to build an online detection model for measurement errors;

[0030] A model training module, configured to train the measurement error online detection model using the training set and the test set;

[0031] A model prediction module, which is used to input the data to be predicted into the trained online measurement error detection model to obtain the corresponding prediction results;

[0032] An error evaluation module is used to perform a measurement error evaluation on the target mutual inductor based on the prediction result; wherein,

[0033] The data to be predicted includes voltage data to be predicted and phase data to be predicted.

[0034] On the basis of the above technical solution, the data set construction module is further used to construct an original data feature table based on historical data;

[0035] The data set construction module is further configured to perform normalization processing on the original data feature table to obtain a processed summary data table;

[0036] The data set construction module is further configured to obtain the voltage history data and the phase history data based on the processed summary data table.

[0037] Based on the above technical solution, the data set construction module is also used to perform missing value processing, outlier processing, duplicate data removal processing and normalization processing on the voltage historical data and the phase historical data to obtain preprocessed voltage historical data and preprocessed phase historical data.

[0038] On the basis of the above technical solution, the model construction module is also used to construct an online detection model for measurement errors based on Transformer and CNN.

[0039] On the basis of the above technical solution, the model building module is further used to extract local features corresponding to the time series through CNN based on the training set and the test set;

[0040] The model building module is also used to perform global information modeling through Transformer based on local features to obtain an online detection model for measurement errors.

[0041] Compared with the prior art, the advantages of the present invention are:

[0042] The present invention can fully utilize the advantages of deep learning technology in feature extraction and pattern recognition, has higher accuracy and robustness, and is suitable for online detection tasks of mutual instrument measurement errors in various industrial and energy fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0044] Figure 1 1 is a flow chart showing the principle of an online detection method for mutual inductor metering error according to an embodiment of the present invention;

[0045] Figure 2 Flowchart of detection steps of a method for online detection of mutual instrument measurement error according to an embodiment of the present invention;

[0046] Figure 3 This is a flow chart of model construction in the online detection method for mutual instrument measurement error according to an embodiment of the present invention;

[0047] Figure 4 Schematic diagram of the model network structure of the online detection method for mutual instrument measurement error according to an embodiment of the present invention. DETAILED DESCRIPTION

[0048] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0049] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0050] The embodiments of the present application provide a method and device for online detection of transformer measurement errors, which can fully utilize the advantages of deep learning technology in feature extraction and pattern recognition, have higher accuracy and robustness, and are suitable for online detection tasks of transformer measurement errors in various industrial and energy fields.

[0051] To achieve the above technical effects, the overall idea of ​​this application is as follows:

[0052] A method for online detection of mutual inductor measurement error, the method comprising the following steps:

[0053] S1. Collect historical data, perform preprocessing, and obtain voltage historical data and phase historical data;

[0054] S2. Preprocess the voltage history data and phase history data, and construct a training set and a test set;

[0055] S3. Construct an online detection model for measurement errors;

[0056] S4. Using the training set and the test set, train an online measurement error detection model;

[0057] S5. Input the data to be predicted into the trained online measurement error detection model to obtain the corresponding prediction result;

[0058] S6. Based on the prediction results, the target mutual inductor is evaluated for measurement error; wherein,

[0059] The data to be predicted includes voltage data to be predicted and phase data to be predicted.

[0060] The embodiments of the present application are further described in detail below with reference to the accompanying drawings.

[0061] First, see Figures 1 to 4As shown, an embodiment of the present application provides an online detection method for mutual inductor measurement error, the method comprising the following steps:

[0062] S1. Collect historical data, perform preprocessing, and obtain voltage historical data and phase historical data;

[0063] S2. Preprocess the voltage history data and phase history data, and construct a training set and a test set;

[0064] S3. Construct an online detection model for measurement errors;

[0065] S4. Using the training set and the test set, train an online measurement error detection model;

[0066] S5. Input the data to be predicted into the trained online measurement error detection model to obtain the corresponding prediction result;

[0067] S6. Based on the prediction results, the target mutual inductor is evaluated for measurement error; wherein,

[0068] The data to be predicted includes voltage data to be predicted and phase data to be predicted.

[0069] In the embodiments of the present application, the advantages of deep learning technology in feature extraction and pattern recognition can be fully utilized, with higher accuracy and robustness, and is suitable for online detection tasks of mutual instrument measurement errors in various industrial and energy fields.

[0070] Furthermore, the preprocessing of the voltage history data and the phase history data and the construction of a training set and a test set includes the following steps:

[0071] Based on historical data, construct the original data feature table;

[0072] Normalizing the original data feature table to obtain a processed summary data table;

[0073] Based on the processed summary data table, the voltage history data and the phase history data are acquired.

[0074] Furthermore, the preprocessing of the voltage history data and the phase history data and the construction of a training set and a test set includes the following steps:

[0075] Missing value processing, outlier processing, duplicate data removal processing, and normalization processing are performed on the voltage history data and the phase history data to obtain preprocessed voltage history data and preprocessed phase history data.

[0076] Furthermore, the construction of the online measurement error detection model includes the following steps:

[0077] Based on Transformer and CNN (Convolutional Neural Networks), an online detection model for measurement errors is constructed.

[0078] Furthermore, the construction of the online measurement error detection model includes the following steps:

[0079] Based on the training set and the test set, extracting local features corresponding to the time series through CNN;

[0080] Based on local features, global information modeling is performed through Transformer to obtain an online detection model for measurement errors.

[0081] Based on the technical solution of the embodiment of the present application, in specific implementation, an online detection system for transformer metering errors based on the feature fusion of Transformer and CNN can be constructed. The system includes a data acquisition module, a feature extraction module, a feature fusion module, a deep learning model and an online detection module. Among them, the data acquisition module is used to collect the signal data of the transformer; the feature extraction module uses the Transformer model and the CNN model to extract features from the signal data; the feature fusion module fuses the features extracted by the Transformer and CNN; the deep learning model uses the fused features for training and learning; the online detection module is used to detect the transformer metering error in real time and output the detection results. Through the feature fusion of Transformer and CNN, the system can make full use of local and global features, thereby improving the accuracy and stability of transformer metering error detection.

[0082] The system and method for online detection of transformer metering errors based on Transformer and CNN feature fusion, provided in the embodiments of this application, can accurately and quickly detect transformer metering errors, improving the reliability and stability of the transformer system. Compared to traditional methods, the embodiments of this application fully utilize the advantages of deep learning technology in feature extraction and pattern recognition, with higher accuracy and robustness, and are suitable for online detection of transformer metering errors in various industrial and energy fields.

[0083] In specific operation, the embodiment of the present application can be an online detection method for mutual inductor measurement error based on Transformer-CNN, and the execution process is as follows:

[0084] Step 1: Collect and store historical data, and select factors that affect measurement error: voltage and phase, that is, obtain voltage historical data and phase historical data;

[0085] Step 2: preprocessing the voltage history data and the phase history data, and obtaining a training set and a test set based on the preprocessed voltage history data and the phase history data;

[0086] Step 3: Establish an online measurement error detection model based on the combination of Transformer and CNN;

[0087] Step 4: Use the training set and the test set to train the measurement error online detection model;

[0088] Step 5: Input the data to be predicted of the same type as the voltage history data and the phase history data into the trained online detection model for metering error, output a prediction result, and judge the trained online detection model for metering error based on the evaluation index.

[0089] In step 1 of the above-mentioned Transformer-CNN-based intelligent short-term power load forecasting method, an original data feature table is established based on the collected voltage history data, the auxiliary features of the data in the original data feature table are annotated, the annotated original data feature table is summarized to establish a summary data table, the feature value of each column in the summary data table is normalized to obtain a processed summary data table, and the processed summary data table is divided into a training set and a test set.

[0090] In step 2 of the above-mentioned online detection method for mutual instrument measurement error based on Transformer-CNN, the data preprocessing includes missing value processing, outlier processing, removal of duplicate data, data cleaning, and normalization of voltage data and feature-affecting data;

[0091] In step 3 of the above-mentioned Transformer-CNN-based online detection method for transformer metering errors, the input signal is first subjected to feature extraction by CNN. CNN captures the local features of the input signal and obtains a series of feature maps. The Transformer model is then used to perform time series modeling on the voltage series information whose features have been extracted by CNN. The voltage data and feature factors are used as input, and the time series is used as output.

[0092] In step 4 of the above-mentioned online detection method for mutual inductor measurement errors based on Transformer-CNN, during the training process, the model can be trained using a training set, and the model parameters can be continuously adjusted to optimize the model.

[0093] Furthermore, the key processes of the embodiments of the present application are described in principle, and the specific contents are as follows:

[0094] First, establish an online detection model for measurement errors, specifically a Transform-CNN combined model

[0095] Assume the original data is Where V t represents the voltage at time t, φ t Represents the phase at time t, and the data set generated by the sliding window is Where L is the window length.

[0096] Extract local features through CNN. Since there are three-phase voltages A, B, and C in a transformer, the voltage and phase of each item can be obtained by measurement. Therefore, the input layer receives data with a shape of (L, 6). After the convolution layer and the pooling layer, the local features of the time series are extracted. Suppose the input data is The convolution kernel is W and the bias is b. The input expression of the convolution layer is:

[0097] H=f(W*X+b) (1)

[0098] Among them, * represents the convolution operation, f represents the activation function (such as RuLE), and finally flattening is performed to flatten the output of the convolution layer into a one-dimensional feature vector.

[0099] The Transformer is used to capture global features. The features extracted by the CNN are used as input to the Transformer to model global information. The CNN output first passes through the Embedding layer, which maps the flattened feature vector to a high-dimensional vector representation. Positional information is then added. The positional encoding formula is as follows:

[0100]

[0101] Where: pos is the position index of the input sequence where the data of a certain time step is located, d model is the dimension of the input sequence word embedding, and i is a dimension of the vector.

[0102] Suppose the output of CNN is The embedded representation is obtained through linear transformation:

[0103] E=HW e +b e (3)

[0104] Among them, W e and b e is the linear transformation parameter. Position encoding is added to preserve sequence information:

[0105] E p =E+P (4)

[0106] Transform is processed through a multi-head attention mechanism and a feedforward neural network:

[0107] Z=MultiHead(E P ) (5)

[0108] O=FFN(Z) (6)

[0109] Among them, MultiHead represents the multi-head attention mechanism, and FFN represents the feedforward neural network. The multi-head attention mechanism is used to capture global features and long-distance dependencies. The self-attention mechanism formula is as follows:

[0110]

[0111] Among them, Q, K, V are the matrices of query, key and value respectively, d model It is the dimension of the key, which can make the gradient more stable during training.

[0112] The feedforward neural network is used to process the features after the self-attention mechanism.

[0113] It's important to note that the term "local feature" is a common term in neural networks and requires no explanation in the patent text. Its meaning is that a local feature focuses on the characteristics of a small portion or local area of ​​data. In an image, this might be a specific corner, edge, or small area of ​​a specific object. Local features are important for identifying or describing specific objects and details in data, as demonstrated in facial recognition, object detection, and feature matching.

[0114] Second, the model predicts the output:

[0115] The output features of CNN and Transform are fused and passed to the fully connected layer for classification or regression. The fused features are expressed as:

[0116] F C =concat(F CNN ,F Transform ) (8)

[0117] Where concat(·) represents the concatenation of output features of different networks, F CNN and F Transform Represent the output features of CNN and Transform network respectively, then use appropriate activation function (such as softmax or sigmoid, etc.), and finally evaluate the output error results.

[0118] Third, model training and evaluation:

[0119] The data is divided into training set, validation set, and test set in a ratio of 8:1:1, and the mean square error (MSE) is used as the loss function.

[0120]

[0121] Among them, u i is the true value of the error, is the output value of the error assessment model, and n represents the sequence length. During the training process, the model performance is evaluated through the validation set, and the model hyperparameters (such as convolution kernel size, pooling window size, etc.) are adjusted to optimize the model performance. Through the above steps, the input data (voltage and phase) undergoes data preprocessing, feature extraction, model training and optimization, and finally an efficient feature extraction model based on CNN is constructed. This model can make full use of the local feature extraction capability of CNN and the global dependency modeling capability of Transformer, and can simultaneously capture the local and global features of the signal, thereby improving the accuracy, robustness and efficiency of detection, and adapting to input data of different types and formats. Compared with using CNN or Transformer alone, the combined model can more effectively handle complex signal patterns and long-term temporal dependencies, resulting in a significant improvement in overall performance.

[0122] In summary, the online detection method and system for mutual instrument measurement errors based on the feature fusion of Transformer and CNN proposed in the embodiments of the present application can fully utilize the feature extraction capabilities of the two models, effectively improving the detection accuracy and robustness of mutual instrument measurement errors.

[0123] By combining the Transformer for global feature extraction and the CNN for local feature extraction, the input data can be understood more comprehensively, and the automated online detection of mutual inductor measurement errors can be achieved through the training and learning of deep learning models. Compared with traditional methods, the online detection method for mutual inductor measurement errors based on the feature fusion of Transformer and CNN proposed in the embodiment of the present application can capture both the global and local features of the data, enabling the model to understand the input data more comprehensively. When processing long sequence data, a single Transformer model may face large computational and memory consumption. By introducing a CNN model for local feature extraction, the length of the input sequence can be reduced, thereby reducing the computational complexity of the model and improving the computational efficiency of the model.

[0124] In a second aspect, an embodiment of the present application provides an online detection device for a mutual inductor measurement error based on the online detection method for a mutual inductor measurement error mentioned in the first aspect, the device comprising:

[0125] The data acquisition module is used to collect historical data, perform preprocessing, and obtain voltage historical data and phase historical data;

[0126] A data set construction module, which is used to preprocess the voltage history data and the phase history data and construct a training set and a test set;

[0127] A model building module, which is used to build an online detection model for measurement errors;

[0128] A model training module, configured to train the measurement error online detection model using the training set and the test set;

[0129] A model prediction module, which is used to input the data to be predicted into the trained online measurement error detection model to obtain the corresponding prediction results;

[0130] An error evaluation module is used to perform a measurement error evaluation on the target mutual inductor based on the prediction result; wherein,

[0131] The data to be predicted includes voltage data to be predicted and phase data to be predicted.

[0132] In the embodiments of the present application, the advantages of deep learning technology in feature extraction and pattern recognition can be fully utilized, with higher accuracy and robustness, and is suitable for online detection tasks of mutual instrument measurement errors in various industrial and energy fields.

[0133] Furthermore, the data set construction module is further used to construct an original data feature table based on historical data;

[0134] The data set construction module is further configured to perform normalization processing on the original data feature table to obtain a processed summary data table;

[0135] The data set construction module is further configured to obtain the voltage history data and the phase history data based on the processed summary data table.

[0136] Furthermore, the data set construction module is also used to perform missing value processing, outlier processing, duplicate data removal processing and normalization processing on the voltage history data and the phase history data to obtain pre-processed voltage history data and pre-processed phase history data.

[0137] Furthermore, the model building module is also used to build an online measurement error detection model based on Transformer and CNN.

[0138] Furthermore, the model building module is further used to extract local features corresponding to the time series through CNN based on the training set and the test set;

[0139] The model building module is also used to perform global information modeling through Transformer based on local features to obtain an online detection model for measurement errors.

[0140] The technical solutions, technical problems solved, and technical effects achieved of the online detection device for mutual inductor metering errors provided in the embodiments of the present application are the same at the technical principle level and will not be elaborated here.

[0141] In the description of this application, it should be noted that the terms "upper" and "lower" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application. Unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be internal communication between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances.

[0142] It should be noted that, in this application, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0143] The foregoing is merely a list of specific embodiments of the present application, intended to enable those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the broadest scope consistent with the principles and novel features of the present application.

Claims

1. A method for online detection of mutual inductor measurement error, characterized in that: The method comprises the following steps: Collect historical data, perform preprocessing, and obtain voltage historical data and phase historical data; Preprocessing the voltage history data and the phase history data, and constructing a training set and a test set; Construct an online detection model for measurement errors; Using the training set and the test set, training the measurement error online detection model; Inputting the data to be predicted into the trained online measurement error detection model to obtain the corresponding prediction result; Based on the prediction results, the target mutual inductor is evaluated for measurement error; wherein, The data to be predicted includes voltage data to be predicted and phase data to be predicted.

2. The online detection method for mutual inductor measurement error according to claim 1, characterized in that: The preprocessing of the voltage history data and the phase history data, and the construction of a training set and a test set, includes the following steps: Based on historical data, construct the original data feature table; Normalizing the original data feature table to obtain a processed summary data table; Based on the processed summary data table, the voltage history data and the phase history data are acquired.

3. The online detection method for mutual inductor measurement error according to claim 1, characterized in that: The preprocessing of the voltage history data and the phase history data, and the construction of a training set and a test set, includes the following steps: Missing value processing, outlier processing, duplicate data removal processing, and normalization processing are performed on the voltage history data and the phase history data to obtain preprocessed voltage history data and preprocessed phase history data.

4. The online detection method for mutual inductor measurement error according to claim 1, characterized in that: The construction of the online measurement error detection model includes the following steps: Based on Transformer and CNN, an online detection model for measurement errors is constructed.

5. The online detection method for mutual inductor measurement error according to claim 4, characterized in that: The construction of the online measurement error detection model includes the following steps: Based on the training set and the test set, extracting local features corresponding to the time series through CNN; Based on local features, global information modeling is performed through Transformer to obtain an online detection model for measurement errors.

6. An online detection device for mutual inductor measurement error, characterized in that: The device comprises: The data acquisition module is used to collect historical data, perform preprocessing, and obtain voltage historical data and phase historical data; A data set construction module, which is used to preprocess the voltage history data and the phase history data and construct a training set and a test set; A model building module, which is used to build an online detection model for measurement errors; A model training module, configured to train the measurement error online detection model using the training set and the test set; A model prediction module, which is used to input the data to be predicted into the trained online measurement error detection model to obtain the corresponding prediction results; An error evaluation module is used to perform a measurement error evaluation on the target mutual inductor based on the prediction result; wherein, The data to be predicted includes voltage data to be predicted and phase data to be predicted.

7. The on-line detection device for mutual inductor measurement error according to claim 6, characterized in that: The data set construction module is further used to construct an original data feature table based on historical data; The data set construction module is further configured to perform normalization processing on the original data feature table to obtain a processed summary data table; The data set construction module is further configured to obtain the voltage history data and the phase history data based on the processed summary data table.

8. The on-line detection device for mutual inductor measurement error according to claim 6, characterized in that: The data set construction module is further used to perform missing value processing, outlier processing, duplicate data removal processing and normalization processing on the voltage history data and the phase history data to obtain pre-processed voltage history data and pre-processed phase history data.

9. The on-line detection device for mutual inductor measurement error according to claim 6, characterized in that: The model building module is also used to build an online measurement error detection model based on Transformer and CNN.

10. The on-line detection device for mutual inductor measurement error according to claim 9, characterized in that: The model building module is further used to extract local features corresponding to the time series through CNN based on the training set and the test set; The model building module is also used to perform global information modeling through Transformer based on local features to obtain an online detection model for measurement errors.

Images