An Adaptive Pooling Method for Dynamic Perception of Power Data Features

Abstract

The present invention belongs to the technical field of data processing, and particularly relates to an adaptive pooling method for dynamic perception of power data features. This method collects multi-source historical and real-time power data, constructs an adaptive pooling model after preprocessing, and the model includes an input layer, a dynamic perception layer, a pooling operation layer, and an output layer. Among them, the dynamic perception layer can dynamically adjust the window size according to data features, and the pooling operation layer adopts a multi-dimensional constraint pooling strategy, including window feature mapping, physical constraint pooling, and feature-sensitive fusion. After the model is trained, it detects real-time power data and compares the detection results with the normal threshold of the power grid to evaluate the operating state. Compared with traditional methods, the present invention can accurately process power data, comprehensively and accurately reflect the operating state of the power grid, effectively solve the deficiencies of traditional technologies in data pooling, noise processing, and comprehensive analysis, and ensure the stable and efficient operation of the power grid.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to an adaptive pooling method for dynamic perception of power data features. Background Art

[0002] In the current situation of the booming development of the power industry, power data processing plays a crucial role in ensuring the stable operation of the power grid and improving the management efficiency of the power system. With the continuous expansion of the power grid scale and the improvement of the intelligent level, power data shows the characteristics of being massive, complex and dynamically changing. Multi-source power data not only contains rich information, but also has different time scales and feature laws. However, existing power data processing technologies have many defects in dealing with these complex data. On the one hand, traditional data pooling methods often use fixed pooling windows and cannot be flexibly adjusted according to the dynamic characteristics of power data, easily losing key information or introducing redundant data, and it is difficult to accurately capture the changing trend of the data. On the other hand, in the data preprocessing link, conventional methods have poor noise processing effects, and may damage the original feature structure of the data while removing noise, affecting the accuracy of subsequent analysis. In addition, existing methods have deficiencies in comprehensively considering the physical characteristics of the power system and fusing multi-source data features, and cannot comprehensively and accurately reflect the actual operating state of the power grid. Summary of the Invention

[0003] In view of the technical problems existing in the above background art, the present invention proposes an adaptive pooling method for dynamic perception of power data features.

[0004] To achieve the above object, the technical solution adopted by the present invention is as follows, including the following steps:

[0005] S1. Data acquisition: Collect multi-source historical power data and multi-source real-time power data; the multi-source historical power data is denoted as the first power data, and the multi-source real-time power data is denoted as the second power data;

[0006] S2. Data preprocessing: Preprocess the collected first power data to obtain the third power data;

[0007] S3. Model construction: Construct an adaptive pooling model;

[0008] The adaptive pooling model includes a data input layer, a dynamic perception layer, a pooling operation layer and an output layer;

[0009] Input layer: Receive the preprocessed power data;

[0010] Dynamic perception layer: Dynamically adjust the perception window size according to the characteristics of the input data;

[0011] Pooling operation layer: Based on the window size output by the dynamic perception layer, a multi-dimensional constrained pooling strategy is adopted;

[0012] Output layer: Output the result after pooling;

[0013] S4. Model training: Input the third power data into the adaptive pooling model for training, and obtain the trained model by continuously adjusting the model parameters;

[0014] S5. Real-time detection: Preprocess the second power data and input it into the trained adaptive pooling model to obtain the detection result;

[0015] S6. Analysis and evaluation: Compare the output result with the normal threshold of the power grid data characteristics to evaluate the operation status of the current power data;

[0016] The multi-dimensional constrained pooling strategy adopted by the pooling operation layer includes window feature mapping, physical constraint pooling, and feature-sensitive fusion;

[0017] Window feature mapping: Map the window size determined by the dynamic perception layer to the pooling dimension parameter;

[0018] Physical constraint pooling: Calculate the constraint violation degree for the power data in each window, set the violation degree threshold, retain the data points greater than the violation degree threshold, and perform average pooling operation on the remaining points;

[0019] Frequency-sensitive fusion: Perform short-time Fourier transform on the pooled data, extract the frequency components, and calculate the weighted sum of the frequency weights to obtain the final feature.

[0020] Preferably, the multi-source historical power data and multi-source real-time power data include power generation data, power transmission data, power distribution data, power consumption data, and power trading data.

[0021] Preferably, the specific implementation of obtaining the third power data is as follows:

[0022] S21. Perform multi-scale decomposition on the first power data The improved wavelet packet basis function satisfies: , where j is the scale factor, k is the translation factor, m is the decomposition branch index, and where , is the traditional basis function, is the improved basis function;

[0023] S22. After wavelet packet transformation, obtain the coefficients of different scales and branches : ;

[0024] S23. Perform soft threshold processing on the wavelet packet data to obtain the processed coefficients , and the calculation method is as follows: ;

[0025] S24. Finally, use the same improved wavelet packet basis function as that in the decomposition to reconstruct the processed wavelet packet coefficients to obtain the third power data , and the calculation method is as follows: .

[0026] Preferably, the calculation method of the soft threshold in step S23 is as follows: , where n is the number of wavelet packet coefficients at this scale and branch, is the noise estimation value at this scale and branch.

[0027] Preferably, between the input layer and the dynamic perception layer of the step S3 model construction, a special extraction operation is required: including data association, feature screening, feature weighting, and feature fusion;

[0028] Feature joint extraction: For the preprocessed power data of the input layer, use a convolutional neural network and a gated recurrent unit to extract spatial features and time series features respectively;

[0029] Feature screening: Calculate the mutual information between features, construct a mutual information matrix, set a mutual information threshold. When the correlation coefficient between two features exceeds this threshold, retain the feature that has a more critical impact on the operation state of the power system and remove redundant features with too high correlation;

[0030] Feature weighting: Use the attention mechanism to assign weights to the screened features, and the weight assignment is dynamically adjusted according to different operation scenarios and data feature importance;

[0031] Feature fusion: Adopt the feature splicing method to fuse different types of features to form a comprehensive feature vector, and then perform dimensionality reduction to retain the key information of the data.

[0032] Preferably, in the dynamic perception layer, the specific steps of dynamically adjusting the perception window size according to the features of the input data include data feature quantization, fluctuation level division, window mapping size, and dynamic adjustment;

[0033] Data feature quantization: Perform quantization analysis on the comprehensive feature vector obtained through the feature extraction operation, calculate the variance of the statistical indicators of each feature, and evaluate the fluctuation degree of the data;

[0034] Fluctuation level division: Set three levels of low fluctuation, medium fluctuation, and high fluctuation, corresponding to different fluctuation ranges respectively, set a threshold and divide according to the fluctuation degree of the data;

[0035] Window size mapping: preset a corresponding perception window size for each fluctuation level;

[0036] Dynamic adjustment: when the fluctuation level of the data changes, adjust the perception window size accordingly.

[0037] Preferably, the violation degree calculation of the physical constraint pooling includes reconstruction error, calculation of residuals, and calculation of violation degree;

[0038] Reconstruction error: construct an autoencoder model, and calculate the Euclidean distance between the original data and the reconstructed data as the reconstruction error;

[0039] Calculation of residuals: based on the admittance matrix of the power grid nodes, calculate the absolute difference between the actual value and the theoretical value of the active power and reactive power as the residuals;

[0040] Calculation of violation degree: based on the weighted fusion of the reconstruction error and the calculated residuals, form a violation degree index.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are that the dynamic perception layer dynamically adjusts the perception window size according to the data characteristics, divides the levels by quantifying the data fluctuations, matches different windows, flexibly adapts to data changes, and avoids information loss or redundancy. The multi-dimensional constraint pooling strategy of the pooling operation layer, combined with window feature mapping, physical constraint pooling, and feature-sensitive fusion, comprehensively considers physical characteristics and frequency characteristics to improve the data processing accuracy. The improved wavelet packet basis function is used to preprocess the data, and combined with multiple algorithms for feature extraction and fusion, effectively suppresses noise, retains key information, and accurately reflects the operation state of the power grid. Description of the Drawings

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0043] Figure 1 Is the overall structural flowchart of the present invention; Figure 2 Is the sub-flowchart of the preprocessing; Figure 3 Is the sub-flowchart of the model construction. Detailed Embodiments

[0044] In order to be able to more clearly understand the above-mentioned objects, features, and advantages of the present invention, the following will further illustrate the present invention in conjunction with the drawings and embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0045] In the following description, many specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those described herein, and thus, the present invention is not limited to the limitations of the specific embodiments disclosed in the following specification.

[0046] Embodiment. In modern power systems, the scale of the power grid continues to expand, the degree of intelligence is constantly improving, the amount of power data is increasing explosively, and it has the characteristics of multi-source, complexity, and dynamic change. Traditional data processing technologies are difficult to adapt to this change, and there are deficiencies in data pooling, noise processing, and comprehensively reflecting the operating state of the power grid, unable to meet the requirements of stable operation and efficient management of the power grid. In order to achieve precise processing of power data, accurately evaluate the operating state of the power grid, ensure the stable and efficient operation of the power grid, and solve the defects of traditional power data processing technologies, the present invention adopts an adaptive pooling method for dynamic perception of power data characteristics, and the specific implementation process is as Figure 1 shown.

[0047] First, sensors and data acquisition devices distributed in various power generation stations, substations, transmission lines, and user terminals are used to collect multi-source historical power data and multi-source real-time power data. The multi-source historical power data (the first power data) covers power generation data over a past period of time, including power generation data, transmission data, distribution data, power consumption data, and power trading data. The multi-source real-time power data (the second power data) is the above-mentioned various types of data collected in real time and is used to reflect the current operating state of the power grid.

[0048] Next, the collected first power data is preprocessed to improve the data quality and obtain the third power data, providing a reliable basis for subsequent analysis.

[0049] Specifically, the first power data is subjected to multi-scale decomposition, and the improved wavelet packet basis function satisfies: , where j is the scale factor, k is the translation factor, m is the decomposition branch index, and among them , is the traditional basis function, is the improved basis function. After wavelet packet transformation, coefficients of different scales and branches are obtained: . The wavelet packet data is subjected to soft threshold processing, and the soft threshold , where n is the number of wavelet packet coefficients of this scale and branch, is the noise estimation value of this scale and branch. The processed coefficients are obtained, and the calculation method is: . Finally, the same improved wavelet packet basis function as that in the decomposition is used to reconstruct the processed wavelet packet coefficients to obtain the third power data , the calculation method is as follows: . After preprocessing, the noise of the data is effectively suppressed, and the feature structure is retained, providing high-quality data for subsequent model training.

[0050] Next, an adaptive pooling model is constructed. The adaptive pooling model is as Figure 3 shown, including a data input layer, a dynamic perception layer, a pooling operation layer, and an output layer. Input layer: Receives the preprocessed power data; Dynamic perception layer: Dynamically adjusts the size of the perception window according to the characteristics of the input data; Pooling operation layer: Adopts a multi-dimensional constraint pooling strategy based on the window size output by the dynamic perception layer; Output layer: Outputs the result after pooling.

[0051] Specifically, the data input layer adopts a parallel processing architecture, which can receive and process data from multiple channels simultaneously, improving data processing efficiency and ensuring the rapid access and transfer of data. The dynamic perception layer first quantifies the data features of the comprehensive power data transmitted from the input layer. For example, calculate the variance of statistical indicators of different types of power data, such as calculating the variance of current data over a period of time to evaluate the degree of data fluctuation. Set three levels of low fluctuation, medium fluctuation, and high fluctuation, corresponding to different variance range thresholds. If the variance of the current data is less than the set low fluctuation threshold, it is determined as the low fluctuation level; if it is within the medium fluctuation threshold range, it is the medium fluctuation level; if it is greater than the high fluctuation threshold, it is the high fluctuation level. A corresponding perception window size is preset for each fluctuation level. For example, the low fluctuation level corresponds to a smaller window, which can finely capture local data features; the high fluctuation level corresponds to a larger window for obtaining more extensive trend information. When the fluctuation level of the data changes, such as from low fluctuation to medium fluctuation, the dynamic perception layer will correspondingly adjust the size of the perception window to adapt to the data change in a timely manner.

[0052] An additional feature extraction operation is also required between the input layer and the dynamic perception layer, as Figure 2 shown: including data combination, feature screening, feature weighting, and feature fusion.

[0053] Feature Joint Extraction: For the preprocessed power data in the input layer, a convolutional neural network and a gated recurrent unit are used to extract spatial features and time series features respectively. Specifically, the preprocessed power data is input into the convolutional neural network CNN. The convolutional layer of CNN slides multiple convolutional kernels over the data to extract local spatial features of the data. Nonlinear factors are introduced through activation functions to enhance the expressive power of the model. The pooling layer downsamples the feature map to reduce the data dimension and improve the computational efficiency. At the same time, the data is input into the gated recurrent unit GRU. GRU controls the flow of information through update gates and reset gates, effectively capturing the time series features of power data, such as the change trend of load over time. Finally, the spatial features extracted by CNN and the time series features extracted by GRU are concatenated to complete the feature joint extraction.

[0054] Feature Screening: Calculate the mutual information between features, construct a mutual information matrix, and set a mutual information threshold. When the correlation coefficient between two features exceeds this threshold, retain the feature that has a more critical impact on the operating state of the power system and remove redundant features with excessive correlation. Specifically, after completing the feature joint extraction, feature screening is carried out for the obtained spatial features and time series features. First, calculate the mutual information for every two features, carefully consider their correlation, and integrate these calculation results to construct a mutual information matrix. Then, according to the characteristics of the power system and actual operation requirements, set a reasonable mutual information threshold. Next, traverse the mutual information matrix to find feature pairs with a correlation coefficient exceeding the threshold. By referring to the operating rules of the power system, historical fault cases, etc., evaluate the impact degree of each feature on the operating state of the power system, retain the more critical feature, and remove redundant features.

[0055] Feature Weighting: Use the attention mechanism to assign weights to the screened features, and the weight assignment is dynamically adjusted according to different operating scenarios and data feature importance. Specifically, in the feature weighting stage, according to the real-time operating parameters of the power system, such as load rate, voltage deviation, equipment health status, etc., construct a risk situation assessment index system. For different operating scenarios, such as peak load, low load, fault repair, etc., determine the weights of each risk index in the current scenario through the fuzzy comprehensive evaluation method. Then, conduct correlation analysis between the screened features and risk indicators, and assign initial weights to the features according to the degree of correlation. During operation, monitor the dynamic changes of the power system in real time. When the risk situation changes, use reinforcement learning algorithms to dynamically adjust the feature weights.

[0056] Feature Fusion: Different types of features are fused by feature concatenation to form a comprehensive feature vector, and then dimensionality reduction is performed to retain the key information of the data. Specifically, different types of features after feature weighting are concatenated in a specific order to construct a comprehensive feature vector. To further optimize the data, a dimensionality reduction method based on quantum-inspired principal component analysis is innovatively adopted. This method introduces the superposition state and entanglement state characteristics of qubits, and on the basis of traditional principal component analysis, more efficiently extracts the main components of the data. First, the comprehensive feature vector is quantum encoded and mapped to the quantum space. Using the parallelism and entanglement characteristics of the quantum state, the projection of the feature vector on the new quantum principal components is quickly calculated. By setting appropriate quantum rotation gate parameters, the extraction direction of the principal components is dynamically adjusted, so as to retain the most critical information for power system analysis to the greatest extent during the dimensionality reduction process.

[0057] Next, in the dynamic perception layer, the specific steps of dynamically adjusting the perception window size according to the characteristics of the input data include data feature quantization, fluctuation level division, window mapping size, and dynamic adjustment. The dynamic perception layer performs quantization analysis on the comprehensive feature vector obtained through feature extraction operations, calculates the variance of the statistical indicators of each feature, and evaluates the fluctuation degree of the data. Three levels of low fluctuation, medium fluctuation, and high fluctuation are set, corresponding to different fluctuation ranges respectively, and the data is divided according to the fluctuation degree. The corresponding perception window size is preset for each fluctuation level. When the fluctuation level of the data changes, the perception window size is adjusted accordingly. For example, when the power load fluctuates slightly within a certain period of time and is at the low fluctuation level, the perception window is set to a smaller value to focus on local details; when the load fluctuation increases and enters the high fluctuation level, the perception window expands to capture a wider range of change trends.

[0058] The pooling operation layer includes adopting a multi-dimensional constraint pooling strategy including window feature mapping, physical constraint pooling, and feature-sensitive fusion.

[0059] Window Feature Mapping: Map the window size determined by the dynamic perception layer to the pooling dimension parameters; specifically, according to the preset rules, map the window time length and width determined by the dynamic perception layer to the step size and kernel size of the pooling operation in the time and feature dimensions respectively.

[0060] Physical Constraint Pooling: Calculate the constraint violation degree for the power data within each window, set a violation degree threshold, retain the data points with a violation degree greater than the threshold, and perform average pooling on the remaining points. Among them, the calculation of the violation degree in physical constraint pooling includes reconstruction error, calculation of residuals, and calculation of violation degree. Reconstruction error: Construct an autoencoder model and calculate the Euclidean distance between the original data and the reconstructed data as the reconstruction error; Calculate residuals: Based on the admittance matrix of the power node, calculate the absolute difference between the actual value and the theoretical value of the active power and reactive power as the residuals; Calculate violation degree: Based on the weighted fusion of the reconstruction error and the calculated residuals, form a violation degree index.

[0061] Frequency-Sensitive Fusion: Perform short-time Fourier transform on the pooled data, extract frequency components, and calculate the weighted sum of the frequency weights to obtain the final feature.

[0062] Real-Time Detection: Preprocess the second power data and input it into the trained adaptive pooling model to obtain the detection result; specifically, first adopt the same preprocessing process as that for the first power data. Subsequently, input the preprocessed second power data into the trained adaptive pooling model. The input layer in the model receives the data, adjusts the window size through the dynamic perception layer, and after being processed by the pooling operation layer, the output layer outputs the detection result, which contains the key feature information of the power data.

[0063] Analysis and Evaluation: Compare the output result with the normal threshold of the power grid data characteristics to evaluate the operating state of the current power data; specifically, compare the detection result output by the model with the pre-set normal threshold of the power grid data characteristics. If it is detected that the current characteristic value of a certain transmission line exceeds the normal current threshold range, or the voltage fluctuation characteristic is abnormal, it is determined that the operating state of the power data of this line is abnormal. Based on these comparison results, the operating state of each part of the current power grid can be comprehensively evaluated, potential fault hazards can be discovered in a timely manner, and strong support can be provided for the power grid operation and maintenance decision-making.

[0064] As described above, it is only a preferred embodiment of the present invention, and it is not a limitation to the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical content of the technical solution of the present invention, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An adaptive pooling method for dynamic perception of power data characteristics, characterized in that: The following steps are involved: S1. Data acquisition: Collect multi-source historical power data and multi-source real-time power data; The multi-source historical power data is recorded as first power data, and the multi-source real-time power data is recorded as second power data; S2, data preprocessing: preprocessing the collected first power data to obtain third power data; S3, model construction: build an adaptive pooling model; The adaptive pooling model includes a data input layer, a dynamic perception layer, a pooling operation layer and an output layer; Input layer: receives pre-processed power data; Dynamic perception layer: dynamically adjusts the perception window size according to the characteristics of the input data; Pooling operation layer: Based on the window size output by the dynamic perception layer, a multi-dimensional constrained pooling strategy is adopted; Output layer: output the result after pooling; S4, model training: inputting the third power data into the adaptive pooling model for training, and obtaining the trained model by continuously adjusting the model parameters; S5, real-time detection: pre-processing the second power data and inputting it into the trained adaptive pooling model to obtain a detection result; S6, analysis and evaluation: compare the output results with the normal threshold of the grid data characteristics to evaluate the operating status of the current power data; The pooling operation layer includes adopting a multi-dimensional constrained pooling strategy including window feature mapping, physical constraint pooling and feature sensitive fusion; Window feature mapping: maps the window size determined by the dynamic perception layer to the pooling dimension parameter; Physical constraint pooling: Calculate the constraint violation degree for the power data in each window, set the violation degree threshold, retain the data points with a value greater than the violation degree threshold, and perform average pooling on the remaining points; Frequency-sensitive fusion: Perform short-time Fourier transform on the pooled data, extract the frequency components, and calculate the weighted sum of the frequency weights to obtain the final features.

2. According to claim 1, an adaptive pooling method for dynamic perception of power data characteristics is characterized in that: The multi-source historical power data and multi-source real-time power data include power generation data, power transmission data, power distribution data, power consumption data, and power transaction data.

3. The adaptive pooling method for dynamic perception of power data characteristics according to claim 1 is characterized in that: The specific implementation of obtaining the third power data is: S21, the first power data Perform multi-scale decomposition and improve the wavelet packet basis function to meet the following requirements: , where j is the scale factor, k is the translation factor, and m is the branch index of the decomposition. , is the traditional basis function, is the improved basis function; S22, after wavelet packet transformation, the coefficients of different scales and branches are obtained : ; S23, perform soft threshold processing on the wavelet packet data to obtain the processed coefficients , calculated as: ; S24, finally use the same improved wavelet packet basis function as in decomposition to process the wavelet packet coefficients Reconstruct and obtain the third power data , calculated as: .

4. The adaptive pooling method for dynamic perception of power data characteristics according to claim 3 is characterized in that: The soft threshold in step S23 The calculation method is: , where n is the number of wavelet packet coefficients of this scale and branch, is the noise estimate for this scale and branch.

5. The adaptive pooling method for dynamic perception of power data characteristics according to claim 1 is characterized in that: The input layer and the dynamic perception layer of the model constructed in step S3 need to undergo special extraction operations: including data union, feature screening, feature weighting and feature fusion; Joint feature extraction: For the pre-processed power data in the input layer, convolutional neural network and gated recurrent unit are used to extract spatial features and time series features respectively; Feature screening: Calculate the mutual information between features, construct a mutual information matrix, and set a mutual information threshold. When the correlation coefficient of two features exceeds the threshold, retain the feature that has a more critical impact on the operating status of the power system and remove redundant features with too high correlation. Feature weighting: Use the attention mechanism to assign weights to the selected features. The weight assignment is dynamically adjusted according to different operating scenarios and the importance of data features. Feature Fusion: The feature splicing method is used to fuse different types of features to form a comprehensive feature vector, and then the dimension is reduced to retain the key information of the data.

6. The adaptive pooling method for dynamic perception of power data characteristics according to claim 1 is characterized in that: In the dynamic perception layer, the specific steps of dynamically adjusting the perception window size according to the characteristics of the input data include data feature quantification, fluctuation level division, window mapping size and dynamic adjustment; Data feature quantification: Quantitatively analyze the comprehensive feature vector obtained through feature extraction, calculate the variance of the statistical index of each feature, and evaluate the degree of data fluctuation; Volatility level classification: set three levels of low volatility, medium volatility and high volatility, corresponding to different volatility ranges, set thresholds and classify according to the degree of data volatility; Window size mapping: pre-set the corresponding perception window size for each fluctuation level; Dynamic adjustment: When the volatility level of the data changes, the perception window size is adjusted accordingly.

7. The adaptive pooling method for dynamic perception of power data characteristics according to claim 1 is characterized in that: The violation degree calculation of the physical constraint pooling includes reconstruction error, calculation residual, and calculation violation degree; Reconstruction error: Build an autoencoder model and calculate the Euclidean distance between the original data and the reconstructed data as the reconstruction error; Calculate the residual: Based on the power node admittance matrix, calculate the absolute difference between the actual value and the theoretical value of the active power and reactive power as the residual; Calculate the violation degree: Based on the weighted fusion of the reconstruction error and the calculation residual, a violation degree indicator is formed.

Images