An intelligent electric quantity anomaly monitoring and missing data completion system

Abstract

This invention relates to the field of power technology and discloses an intelligent power anomaly monitoring and missing data completion system. This system includes a data preprocessing module, a customized attention model, and prediction and processing modules. Through a customized attention mechanism, the model can more accurately capture complex relationships between data, especially dependencies across tables and fields, thereby improving the accuracy and reliability of missing value prediction. By combining business knowledge and statistical information into the model, and through field relationship graph construction and attention weight adjustment, the prediction results are more consistent with actual business scenarios. When using this model to process large-scale multi-table data, it can maintain high processing speed and efficiency, reducing computational and time costs. This technical solution demonstrates significant advantages in improving prediction accuracy, enhancing business adaptability, and optimizing computational efficiency.

Description

Technical Field

[0001] This invention belongs to the field of power technology, specifically a smart power anomaly monitoring and missing data completion system. Background Technology

[0002] In electricity market transactions, existing technologies have proposed a series of data fitting rules to address the issue of missing data caused by meter data collection failures. These rules include adopting different fitting strategies based on the number of defects within a continuous time period. For example, when the number of defects is small, the arithmetic mean of the electricity consumption between consecutive time points is used for fitting; when the number of defects is large, corresponding historical data is selected for approximate fitting based on date attributes. These techniques provide an effective solution for supplementing missing data when meter collection fails.

[0003] In the field of power data acquisition, existing power acquisition terminals often miss some data when reading meter data due to network fluctuations, equipment failures, and other reasons. In particular, the omission of critical information such as daily frozen data and curve data has a serious impact on power grid monitoring and user assessment. Traditional methods rely heavily on manual verification and manual data entry, which is not only inefficient but also prone to errors. In recent years, with the development of broadband carrier technology, the success rate of automatic meter reading by power acquisition terminals has significantly improved. However, the problem of missed readings due to the inability to collect curve data during the daily frozen data reading period still exists. Therefore, an efficient and automated data supplementation technology is needed to ensure that power acquisition terminals can obtain all meter data completely and accurately, thereby meeting the needs of power grid operation and management. The present invention is proposed to solve this technical problem. By referring to the correlation between other fields and missing fields, an intelligent algorithm is used to achieve automatic data completion, improving data integrity and accuracy, and thus improving the efficiency and intelligence level of power grid operation and management. Summary of the Invention

[0004] To address the above issues and overcome the shortcomings of existing technologies, this invention provides an intelligent power anomaly monitoring and missing data completion system. This system effectively solves the problem that in the field of power data acquisition, existing power acquisition terminals often miss some data when reading meter data due to network fluctuations, equipment failures, and other reasons. In particular, the omission of critical information such as daily frozen data and curve data has a serious impact on power grid monitoring and user assessment. Traditional methods rely heavily on manual verification and data entry, which is not only inefficient but also prone to errors. In recent years, with the development of broadband carrier technology, the success rate of automatic meter reading by power acquisition terminals has significantly improved, but the problem of missed readings due to the inability to collect curve data during the daily frozen data reading period still exists.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent power anomaly monitoring and missing data completion system, the intelligent power anomaly monitoring and missing data completion system comprising a data preprocessing module, a customized attention model, and a prediction and processing module;

[0006] Data preprocessing module: The data preprocessing module includes a data cleaner, a serializer, and a table embedding generator; the data cleaner is responsible for identifying and processing outliers and missing values ​​in the input multi-table data, pre-filling missing values, and data normalization. The pre-filling method for missing values ​​includes filling missing values ​​with zero, mean, median, mode, or combined with specific business requirements; the serializer sorts and organizes the data according to preset rules; the table embedding generator is responsible for capturing table-level information, and the system generates unique embedding vectors through the tables;

[0007] Customized Attention Model: The customized attention model includes a basic attention layer and a table relationship enhancement layer. The basic attention layer uses a self-attention mechanism to process the serialized data, which can capture complex dependencies between elements within the sequence. The table relationship enhancement layer includes a field relationship graph builder, an attention weight adjuster, and a cross-table attention mechanism. The field relationship graph builder is based on business knowledge and statistical information, constructing a field relationship graph with fields as nodes and dependencies or correlations between fields as edges. The attention weight adjuster uses table embeddings and the field relationship graph to adjust the attention weights calculated by the basic attention layer. The cross-table attention mechanism allows the model to consider information from different tables when calculating attention weights.

[0008] Prediction and Processing Module: The prediction and processing module includes a prediction network and a processing module. The prediction network is responsible for predicting the value of the missing field using enhanced attention weights and other field values ​​in the sequence. The prediction network adopts a fully connected layer deep learning architecture. The processing module includes an evaluation metric and an output formatter. The evaluation metric is used to measure the degree of difference between the model's prediction results and the actual results, as well as the overall performance of the model. The output formatter is used to ensure that the prediction results can be seamlessly integrated into existing data systems. The formatter performs data cleaning and organization as needed.

[0009] Preferably, the outlier detection in the data preprocessing module uses the Z-score statistical method, and the Z-score calculation formula is as follows: The data preprocessing formula for the data cleaner is:

[0010] Preferably, the basic attention layer adopts the multi-head attention mechanism in the Transformer model, and the self-attention calculation formula for each group of independent heads in the multi-head attention mechanism is as follows: Multi-head attention performs multiple independent attention operations in parallel and then concatenates the results. The concatenation formula is MultiHead(Q, K, V) = Concat(head1, ..., head2, ..., head3, ..., head4, ..., head5, ..., head6, ..., head7, ..., head8, ..., head9, ..., head1 ... h W 0 The attention weight adjuster uses table embeddings and field relationship graphs to adjust the attention weights generated by the base attention layer. The adjusted attention weights are represented as W. adj =σ(W atta +Weightfunction(A, T)); Cross-table attention mechanism allows the model to consider information from different tables when calculating attention weights by introducing cross-table embeddings or relation matrices. The cross-table attention weights are designed as follows:

[0011] Preferably, the prediction network in the prediction and processing module adopts a multi-layer fully connected neural network, which includes an input layer, a hidden layer, and an output layer.

[0012] Preferably, during the prediction process, the input layer ignores pre-filled default values. The prediction network employs a masking mechanism. The input layer receives feature vectors processed by the customized attention model, which include serialized data field values ​​and a mask. The input layer formula is as follows:

[0013] Preferably, the hidden layer adopts a multi-layer fully connected neural network structure, and each fully connected neural network layer uses a non-linear activation function to increase the non-linearity of the model, which is expressed by the formula h. l =ReLU(W l h l-1 +b l ).

[0014] Preferably, the output layer is designed according to a specific task, and its output prediction formula is as follows:

[0015] Preferably, the processing module is responsible for evaluating the quality of the prediction results and preparing output data for subsequent use, using evaluation metrics MSE, MAE, and RMSE to measure the accuracy of the model predictions, wherein the formula for MSE is... The formula for MAE is: The formula for RMSE is:

[0016] Preferably, the output formatter is used to organize the prediction results into a format that meets business requirements, including filling the prediction values ​​back into the corresponding positions of the original data table, performing necessary data cleaning and organization work, generating reports or visualization charts, and integrating the prediction results into the data system.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. The present invention, through a customized attention mechanism, enables the model to more accurately capture the complex relationships between data, especially the dependencies between tables and fields, thereby improving the accuracy and reliability of missing value prediction;

[0018] 2. This invention combines business knowledge and statistical information into the model, and through the construction of field relationship graphs and adjustment of attention weights, makes the prediction results more consistent with actual business scenarios;

[0019] 3. When using this model to process large-scale multi-table data, the present invention can maintain high processing speed and efficiency, and reduce computing and time costs. This technical solution has shown significant advantages in improving prediction accuracy, enhancing business adaptability, and optimizing computing efficiency. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0021] In the attached diagram:

[0022] Figure 1 This is a block diagram of an intelligent power anomaly monitoring and missing data completion system according to the present invention. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0024] like Figure 1 As shown, the present invention provides a technical solution: an intelligent power anomaly monitoring and missing data completion system, which includes a data preprocessing module, a customized attention model, and a prediction and processing module;

[0025] Data preprocessing module: The data preprocessing module includes a data cleaner, a serializer, and a table embedding generator; the data cleaner is responsible for identifying and processing outliers and missing values ​​in the input multi-table data, pre-filling missing values, and data normalization. The pre-filling method for missing values ​​includes filling missing values ​​with zero, mean, median, mode, or combined with specific business requirements; the serializer sorts and organizes the data according to preset rules; the table embedding generator is responsible for capturing table-level information, and the system generates unique embedding vectors through the tables;

[0026] Customized Attention Model: The customized attention model includes a basic attention layer and a table relationship enhancement layer. The basic attention layer uses a self-attention mechanism to process the serialized data, which can capture complex dependencies between elements within the sequence. The table relationship enhancement layer includes a field relationship graph builder, an attention weight adjuster, and a cross-table attention mechanism. The field relationship graph builder is based on business knowledge and statistical information, constructing a field relationship graph with fields as nodes and dependencies or correlations between fields as edges. The attention weight adjuster uses table embeddings and the field relationship graph to adjust the attention weights calculated by the basic attention layer. The cross-table attention mechanism allows the model to consider information from different tables when calculating attention weights.

[0027] Prediction and Processing Module: The prediction and processing module includes a prediction network and a processing module. The prediction network is responsible for predicting the value of the missing field using enhanced attention weights and other field values ​​in the sequence. The prediction network adopts a fully connected layer deep learning architecture. The processing module includes an evaluation metric and an output formatter. The evaluation metric is used to measure the degree of difference between the model's prediction results and the actual results, as well as the overall performance of the model. The output formatter is used to ensure that the prediction results can be seamlessly integrated into existing data systems. The formatter performs data cleaning and organization as needed.

[0028] The outlier detection in the data preprocessing module uses the Z-score statistical method, and the Z-score calculation formula is as follows: The data preprocessing formula for the data cleaner is: The base attention layer employs the multi-head attention mechanism from the Transformer model. The self-attention calculation formula for each independent head in the multi-head attention mechanism is as follows: Multi-head attention performs multiple independent attention operations in parallel and then concatenates the results. The concatenation formula is MultiHead(Q, K, V) = Concat(head1, ..., head2, ..., head3, ..., head4, ..., head5, ..., head6, ..., head7, ..., head8, ..., head9, ..., head1 ... h W 0 The attention weight adjuster uses table embeddings and field relationship graphs to adjust the attention weights generated by the base attention layer. The adjusted attention weights are represented as W. adj =σ(W atta +Weightfunction(A, T)); Cross-table attention mechanism allows the model to consider information from different tables when calculating attention weights by introducing cross-table embeddings or relation matrices. The cross-table attention weights are designed as follows: The prediction network in the prediction and processing module employs a multi-layer fully connected neural network, which includes an input layer, hidden layers, and an output layer. The input layer is used during prediction, ignoring pre-filled default values. The prediction network uses a masking mechanism. The input layer receives the feature vector processed by the customized attention model, which includes serialized data field values ​​and a mask. The input layer formula is as follows: The hidden layers employ a multi-layer fully connected neural network structure. Each fully connected neural network layer uses a non-linear activation function to increase the model's non-linearity, expressed as h. l =ReLU(W l h l-1 +b l The output layer is designed according to the specific task, and its output prediction formula is as follows: The processing module is responsible for evaluating the quality of the prediction results and preparing output data for subsequent use. It uses evaluation metrics MSE, MAE, and RMSE to measure the accuracy of the model predictions. The formula for MSE is: The formula for MAE is: The formula for RMSE is: Output formatters are used to format forecast results into a format that meets business needs, including filling forecast values ​​back into the corresponding positions in the original data tables, performing necessary data cleaning and processing, generating reports or visualizations, and integrating forecast results into the data system.

[0029] The data preprocessing module ensures that subsequent models receive clean, ordered, and structured data input. The data cleaner within the preprocessing module identifies and processes outliers in the multi-table input data, performs default value pre-filling, and normalizes the data. Outliers may include values ​​outside the reasonable range or obviously erroneous data points. Before inputting the sequence into the model, missing values ​​are pre-filled using methods such as zero, mean, median, mode, or based on specific business requirements. Data normalization reduces data redundancy and improves data consistency, enhancing data accuracy and completeness, thereby improving performance. For query efficiency and ease of data management, the serializer is used to further serialize the cleaned data so that subsequent models can process it efficiently. The serializer sorts and organizes the data according to preset rules to ensure data consistency and orderliness, which helps the model better understand the relationships and dependencies between data. The table embedding generator is used to capture table-level information. The system generates a unique embedding vector for each table. These embedding vectors are obtained through a business knowledge-based initialization method and can reflect the structure, content, and position and role of the table in multi-table data. These embedding vectors will serve as one of the important inputs to the customized attention model.

[0030] The base attention layer in the customized attention model employs a self-attention mechanism to process the serialized data. This mechanism captures complex dependencies between elements within a sequence, which is crucial for understanding the interactions between different fields in table data. The table relationship enhancement layer includes a field relationship graph builder, an attention weight adjuster, and a cross-table attention mechanism. The field relationship graph builder, based on business knowledge and statistical information, constructs a field relationship graph with fields as nodes and dependencies or correlations between fields as edges. This graph structure helps the model understand the complex relationships between fields and adjust attention weights accordingly. The attention weight adjuster adjusts the attention weights calculated by the base attention layer using table embeddings and the field relationship graph. By introducing a learnable weight matrix or function, the model can learn how to optimize the allocation of attention weights based on table embeddings and the field relationship graph. This allows the model to focus more on information and relationships related to missing fields when predicting missing values. The cross-table attention mechanism further enhances the model's cross-table learning ability by allowing the model to consider information from different tables when calculating attention weights. By incorporating cross-table embedding or relational information, the model can gain a more comprehensive understanding of the complex relationships between data from multiple tables and make more accurate predictions of missing values.

[0031] The prediction network in the prediction and processing module is one of the core components of the system. It is responsible for predicting the values ​​of missing fields using enhanced attention weights and other field values ​​in the sequence. The prediction network can employ a fully connected deep learning architecture. During prediction, the network uses a masking mechanism to distinguish between non-missing values ​​and default values. This aims to minimize the impact of pre-filled default values ​​during the prediction of missing values, focusing instead on non-default values. An appropriate loss function is used for back-derivative calculation to derive the most suitable parameters. The processing module includes evaluation metrics and an output formatter. Evaluation metrics measure the difference between the model's predictions and actual results, as well as the overall performance of the model. These metrics are crucial for understanding the model's accuracy, bias, and variance. The output formatter ensures that the prediction results can be seamlessly integrated into existing data systems for easy subsequent use and analysis. The formatter can also perform data cleaning and preparation as needed to ensure the quality and consistency of the output data.

[0032] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0033] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart power anomaly monitoring and missing data completion system, characterized in that: The intelligent power anomaly monitoring and missing data completion system includes a data preprocessing module, a customized attention model, and a prediction and processing module; Data preprocessing module: The data preprocessing module includes a data cleaner, a serializer, and a table embedding generator; The data cleaner is responsible for identifying and processing outliers and default values ​​in the input multi-table data, as well as pre-filling and normalizing the data. The default value pre-filling method includes using zero, mean, median, mode, or filling default values ​​according to specific business requirements. The serializer sorts and organizes the data according to preset rules. The table embedding generator is responsible for capturing table-level information, and the system generates unique embedding vectors through the tables. Customized attention model: The customized attention model includes a basic attention layer and a table relationship enhancement layer; the basic attention layer uses a self-attention mechanism to process the serialized data, and the self-attention mechanism can capture the complex dependencies between elements within the sequence; The table relationship enhancement layer includes a field relationship graph builder, an attention weight adjuster, and a cross-table attention mechanism. The field relationship graph builder constructs a field relationship graph based on business knowledge and statistical information, with fields as nodes and dependencies or correlations between fields as edges. The attention weight adjuster uses table embeddings and field relationship graphs to adjust the attention weights calculated by the base attention layer; Cross-table attention mechanism allows the model to consider information from different tables when calculating attention weights. The base attention layer adopts the multi-head attention mechanism in the Transformer model. The self-attention calculation formula for each group of independent heads in the multi-head attention mechanism is as follows: Multi-head attention performs multiple independent attention operations in parallel and then concatenates the results. The concatenation formula is MultiHead(Q,K,V)=Concat(head1,...,head) h W 0 The attention weight adjuster uses table embeddings and field relationship graphs to adjust the attention weights generated by the base attention layer. The adjusted attention weights are represented as W. adj =σ(W atta +Weightfunction(A,T); Cross-table attention mechanisms allow the model to consider information from different tables when calculating attention weights by introducing cross-table embeddings or relation matrices. The cross-table attention weights are designed as follows: Prediction and Processing Module: The prediction and processing module includes a prediction network and a processing module. The prediction network is responsible for predicting the value of the missing field using enhanced attention weights and other field values ​​in the sequence. The prediction network adopts a fully connected layer deep learning architecture. The processing module includes an evaluation metric and an output formatter. The evaluation metric is used to measure the degree of difference between the model's prediction results and the actual results, as well as the overall performance of the model. The output formatter is used to ensure that the prediction results can be seamlessly integrated into existing data systems. The formatter performs data cleaning and organization tasks as needed.

2. The intelligent power anomaly monitoring and missing data completion system according to claim 1, characterized in that: The prediction network in the prediction and processing module adopts a multi-layer fully connected neural network, which includes an input layer, a hidden layer, and an output layer.

3. The intelligent power anomaly monitoring and missing data completion system according to claim 2, characterized in that: The input layer is used to ignore pre-filled default values ​​during the prediction process. The prediction network employs a masking mechanism. The input layer receives the feature vector processed by the customized attention model, which includes serialized data field values ​​and a mask. The input layer formula is X. out =Attention(Q,K,V)⊕X mask .

4. The intelligent power anomaly monitoring and missing data completion system according to claim 3, characterized in that: The hidden layer employs a multi-layer fully connected neural network structure. Each fully connected neural network layer uses a non-linear activation function to increase the model's non-linearity, expressed as h. l =ReLU(W l h l-1 +b l ).

5. The intelligent power anomaly monitoring and missing data completion system according to claim 4, characterized in that: The output layer is designed according to the specific task, and its output prediction formula is as follows:

6. The intelligent power anomaly monitoring and missing data completion system according to claim 1, characterized in that: The processing module is responsible for evaluating the quality of the prediction results and preparing output data for subsequent use. It uses evaluation metrics MSE, MAE, and RMSE to measure the accuracy of the model predictions, where the formula for MSE is... The formula for MAE is: The formula for RMSE is:

7. The intelligent power anomaly monitoring and missing data completion system according to claim 1, characterized in that: The output formatter is used to organize the prediction results into a format that meets business needs, including filling the prediction values ​​back into the corresponding positions of the original data table, performing necessary data cleaning and organization work, generating reports or visualization charts, and integrating the prediction results into the data system.

Images