Real-time metering method and system of intelligent modular electric energy metering box

Abstract

The invention relates to the field of electric energy metering, and discloses a real-time metering method of an intelligent modular electric energy metering box, which comprises the following steps: deploying a sensor array in the electric energy metering box; acquiring original electric energy data and environmental parameters through the sensor array to obtain an original electric energy metering data stream; performing dynamic threshold filtering on the original electric energy metering data stream to obtain clean electric energy metering data; constructing an electric energy topology model according to a graph neural network, and inputting the key electric energy parameters into the electric energy topology model to obtain initial electric energy topology network data; and abnormal nodes in the electric energy topology model are detected through an isolated forest algorithm to obtain optimized electric energy topology data, the dynamic threshold filtering and Kalman filtering algorithms are adopted, noise and abnormal values in the original electric energy data can be effectively removed, the reliability of the electric energy metering data is ensured, and especially in a complex environment, the reliability of the electric energy metering data is improved. Errors may be caused by factors such as electromagnetic interference.

Description

Technical Field

[0001] The present invention relates to the technical field of electric energy metering, and in particular to a real-time metering method and system for an intelligent modular electric energy metering box. Background Art

[0002] The meter box is a set of metering instruments and auxiliary equipment necessary for measuring electric energy, including electric energy meters, voltage and current transformers and their secondary circuits, electric energy metering panels, cabinets, boxes, etc.

[0003] Existing equipment has the following shortcomings: Traditional methods often use simple filters or fixed thresholds to remove noise when processing power data. This may not effectively cope with noise and outliers in complex environments, such as electromagnetic interference, resulting in inaccurate and unreliable power metering data. Traditional methods cannot dynamically adjust thresholds, making it difficult to adapt to environmental changes and may ignore or incorrectly process some outliers. Traditional methods typically use fixed algorithms to analyze grid topology, lacking the ability to accurately identify potential grid problems. Fault points or unreasonable power distribution in the grid may be missed, making it difficult to detect potential grid problems in a timely manner.

[0004] Therefore, the present application proposes a real-time metering method and system for an intelligent modular electric energy metering box to solve the above-mentioned problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a real-time metering method and system for an intelligent modular electricity metering box to solve the problem that the above-mentioned traditional methods often use simple filters or fixed thresholds to remove noise when processing electricity data. This may not be able to effectively deal with noise and outliers in complex environments, such as electromagnetic interference and other factors, resulting in inaccurate and unreliable electricity metering data.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a real-time metering method for an intelligent modular electric energy metering box, comprising: A sensor array is deployed inside the electric energy meter box; raw electric energy data and environmental parameters are collected by the sensor array to obtain a raw electric energy metering data stream; and dynamic threshold filtering is performed on the raw electric energy metering data stream to obtain clean electric energy metering data; Performing principal component analysis on the clean electric energy metering data to obtain key electric energy parameters; constructing an electric energy topology model based on a graph neural network, inputting the key electric energy parameters into the electric energy topology model to obtain initial electric energy topology network data; detecting abnormal nodes in the electric energy topology model using an isolation forest algorithm to obtain optimized electric energy topology data; Acquire environmental data, and compensate the optimized power topology data for environmental effects based on the environmental data to obtain compensated optimized power topology data; dynamically correct the clean power metering data based on the compensated optimized power topology data using a Kalman filter algorithm to obtain actual power metering data; and map the actual power metering data to standard power units using an edge computing node to obtain standardized power metering data; The standardized electricity metering data is input into a pre-trained fusion model. The fusion model uses LSTM and CNN to extract time series features and fuse them with spatial features to obtain electricity usage pattern labels. The electricity usage pattern labels are compared with preset electricity usage specification data to calculate the electricity usage compliance score to obtain a smart metering report. The smart metering report includes real-time metering values, electricity usage patterns, and compliance scores.

[0007] A sensor array is deployed inside the electric energy meter box; the sensor array collects raw electric energy data and environmental parameters to obtain a raw electric energy metering data stream, including: Design a modular sensor array architecture inside the electric energy meter box, the sensor array including high-precision current transformers, voltage sensors, and temperature and humidity sensors; An ultra-low latency communication link is established through the network, raw power data and environmental parameters are periodically collected, and timestamps and device ID tags are added to obtain the raw power metering data stream.

[0008] The raw data is dynamically filtered using a threshold value to obtain clean energy metering data, including: A dynamic threshold benchmark library is built based on historical power data. The sliding window algorithm is used to calculate the mean and standard deviation of the data within the window to obtain the adaptive threshold range. An environmental interference model is established based on temperature and humidity sensor data to obtain the environmental adaptive threshold parameters. Perform real-time deviation detection on the raw power data stream and calculate the absolute error between the current data point and the sliding window mean. Use the soft threshold shrinkage method to denoise the high-frequency noise band to obtain preliminary denoised data. Cyclic redundancy check is used to detect the integrity of the data packet to obtain the verified intermediate data; The verified intermediate data is input into the improved Kalman filter, and state estimation is performed in combination with dynamic threshold parameters and wavelet denoising results; data deviation is corrected through a prediction update cycle to obtain clean electricity metering data.

[0009] The clean electric energy metering data is subjected to principal component analysis to identify key electric energy parameters; an electric energy topology model is constructed based on a graph neural network to associate the electric energy data of each loop to obtain initial electric energy topology network data, including: The clean electric energy metering data is standardized, and the principal components are screened by cumulative variance contribution rate to obtain a reduced dimension feature matrix; the reduced dimension feature matrix includes harmonic content, three-phase imbalance and power factor; A multimodal graph structure is constructed, and each circuit in the electric energy meter box is defined as a graph node. The correlation between the electric energy parameters of the graph nodes is calculated using the Pearson correlation coefficient to obtain an initial electric energy topology graph. Deploy a graph attention network and perform multi-layer graph convolution operations on the initial topology graph; dynamically adjust edge weights through the attention mechanism to obtain optimized power topology network data.

[0010] Among them, the isolation forest algorithm is used to detect abnormal nodes in the topological network and optimize the power metering path to obtain optimized power topology data, including: Extract node feature vectors from the initial power topology network data; construct a node feature matrix and retain edge connection relationships to obtain topological feature data suitable for isolation forest input; An isolation forest model is constructed, and a plurality of isolation trees are constructed by randomly sampling subspaces on the node feature matrix; an anomaly score is calculated by path length to obtain an anomaly detection result; Optimizing the topology network based on the anomaly detection results and removing abnormal nodes; calculating the shortest path of the topology network optimized by the anomaly detection results in combination with the efficiency requirements of the electric energy metering path to obtain an optimized edge weight matrix and node list; The optimization effect is verified by simulating the injection of abnormal data, and the anomaly detection accuracy and path delay are calculated to obtain the optimized power topology network data.

[0011] The optimized power topology data is compensated for environmental effects, and an electromagnetic interference compensation model is established in combination with the temperature and humidity sensor data. The metering data is dynamically corrected using the Kalman filter algorithm to obtain the actual power metering value after compensation: Extracting node features and topological attributes from the optimized power topology data; collecting temperature and humidity sensor data to construct a multimodal data set including power parameters and environmental parameters to obtain an environmental power correlation feature matrix; An electromagnetic interference compensation model is constructed based on a graph attention network, and the nonlinear effect of the environment on the multimodal dataset is learned through multi-layer graph convolution to obtain node-level electromagnetic interference compensation coefficients; Applying the node-level electromagnetic interference compensation coefficient to the optimized power topology data, performing reverse correction on parameters such as voltage and current to obtain preliminary corrected data after compensation; deploying edge computing nodes to perform Kalman filtering, using the compensation data as observation values ​​and the model prediction value as state estimation, and dynamically correcting the noise through a prediction update loop to obtain an actual power metering value with a noise level below a threshold; The compensation effect is verified by simulating environmental change experiments, and the data errors of the actual electric energy measurement values ​​before and after correction are calculated to obtain the actual electric energy measurement values ​​after compensation.

[0012] The edge computing nodes are used to map the actual measurement values ​​to standard electric energy units to obtain standardized electric energy measurement data, including: Deploy a unit conversion engine at the edge computing node to receive the actual compensated energy metering value. Use a pre-trained neural network model to identify the energy parameter type to obtain initial standardized data. Deploy a dynamic calibration strategy, combine the historical data of the standard electric energy meter stored in the edge node, calculate the deviation between the initial standardized data and the standard value, and obtain intermediate standardized data; A timing model within the edge computing node is used to perform nonlinear error correction on the intermediate standardized data; by capturing the time correlation of the electric energy parameters, the accumulated error in the unit conversion process is predicted and corrected to obtain standardized electric energy metering data; The standardized electricity metering data is synchronized to the cloud blockchain platform, and a data fingerprint is generated through a hash algorithm and stored on the chain to obtain a standardized electricity metering data stream.

[0013] The standardized electric energy metering data is input into a pre-trained fusion model, and the electric energy usage pattern label is obtained by extracting temporal features and fusing spatial features, including: The standardized electric energy metering data is divided into multi-scale time windows to obtain short-time, medium-time, and long-time time series segments; the frequency domain features of each segment are extracted by wavelet transform, and a multimodal feature matrix is ​​constructed by combining time domain statistics to obtain enhanced data; A parallel LSTM and CNN architecture is deployed. The LSTM branch captures the dynamic changes of the enhanced data, and the CNN branch extracts the spatial distribution pattern of frequency domain features. The outputs of the two branches are fused through a cross-attention mechanism to obtain a fused feature vector.

[0014] The standardized electric energy metering data is input into a pre-trained fusion model, and the electric energy usage pattern label is obtained by extracting time series features and fusing them with spatial features. The method further includes: Inputting the fused feature vector into a pre-trained classification layer and combining it with a power usage scenario label library for pattern matching; dynamically adjusting the classification threshold using a reinforcement learning algorithm to obtain a power usage pattern label; The accuracy of labels is verified by back-tracing historical power data, and the classification F1 value is calculated to obtain the power usage pattern label flow.

[0015] The real-time metering system of the intelligent modular electric energy metering box is applicable to the real-time metering method of the intelligent modular electric energy metering box, including: A data filtering unit is configured to deploy a sensor array inside the electric energy metering box; collect raw electric energy data and environmental parameters through the sensor array to obtain a raw electric energy metering data stream; and perform dynamic threshold filtering on the raw electric energy metering data stream to obtain clean electric energy metering data; an electric energy topology unit, performing principal component analysis on the clean electric energy metering data to obtain key electric energy parameters; constructing an electric energy topology model based on a graph neural network, inputting the key electric energy parameters into the electric energy topology model to obtain initial electric energy topology network data; and detecting abnormal nodes in the electric energy topology model using an isolation forest algorithm to obtain optimized electric energy topology data; a data correction unit, which acquires environmental data, compensates the optimized power topology data for environmental effects based on the environmental data to obtain compensated optimized power topology data; dynamically corrects the clean power metering data based on the compensated optimized power topology data using a Kalman filter algorithm to obtain actual power metering data; and maps the actual power metering data to standard power units using an edge computing node to obtain standardized power metering data; The electricity usage specification unit inputs the standardized electricity metering data into a pre-trained fusion model, extracts time series features and fuses them with spatial features to obtain an electricity usage pattern label; compares the electricity usage pattern label with preset electricity usage specification data, calculates the electricity usage compliance score, and obtains a smart metering report; wherein the smart metering report includes real-time metering values, electricity usage patterns, and compliance scores.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention can effectively remove noise and outliers from raw power data by adopting dynamic threshold filtering and Kalman filtering algorithms, thereby improving the accuracy of power metering. This ensures the reliability of power metering data, especially in complex environments where errors may be caused by factors such as electromagnetic interference. By using a graph neural network to construct an energy topology model and employing the isolation forest algorithm to detect abnormal nodes in the topological network, this method can accurately identify potential problems in the power grid, such as fault points or unreasonable energy distribution. This provides strong support for power grid optimization and fault diagnosis, helping to promptly discover and resolve problems in the power transmission process. This invention combines temperature and humidity sensor data with an electromagnetic interference compensation model to effectively address the impact of environmental changes on energy metering. For example, temperature and humidity changes can affect the accuracy of equipment. Correction through the compensation model ensures that measurement results are not affected by external environmental fluctuations, thereby improving metering stability and reliability. This paper extracts key parameters from power data through principal component analysis and combines it with an LSTM-CNN fusion model to not only measure power but also extract temporal and spatial features, helping to build more accurate power usage patterns. The generation of power usage pattern labels facilitates further analysis and prediction of device usage behavior. By comparing power usage pattern tags with pre-set power usage specification data, the present invention can generate power usage compliance scores and automatically generate smart metering reports. This can provide real-time compliance detection for users or operators, helping to promptly identify non-compliant behaviors in power usage and promote energy conservation and emission reduction. By utilizing edge computing nodes to process and transmit data, the present invention enables electricity metering and analysis to be completed quickly locally, reducing dependence on central servers and improving the system's response speed and processing efficiency. This is especially important for electricity management systems that require real-time feedback. When problems occur in equipment or the power grid, it can immediately respond and issue an alarm, monitor anomalies in electricity metering data and power topology in real time, and the system can proactively issue fault warnings, promptly repair potential faults, or optimize maintenance strategies, thereby reducing power grid downtime and fault repair costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the steps of the overall method in one embodiment of the present invention; Figure 2 FIG. 1 is a schematic diagram of the system architecture structure of the overall system in one embodiment of the present invention.

[0018] In the figure: 1. Data filtering unit; 2. Power topology unit; 3. Data correction unit; 4. Power usage specification unit. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0020] See also Figure 1-2 The present invention provides a technical solution: a real-time metering method for an intelligent modular electric energy metering box, comprising: S1. Deploy a sensor array inside an electric energy meter box; collect raw electric energy data and environmental parameters through the sensor array to obtain a raw electric energy metering data stream; and perform dynamic threshold filtering on the raw electric energy metering data stream to obtain clean electric energy metering data. S2. Performing principal component analysis on the clean electric energy metering data to obtain key electric energy parameters; constructing an electric energy topology model based on a graph neural network, inputting the key electric energy parameters into the electric energy topology model to obtain initial electric energy topology network data; detecting abnormal nodes in the electric energy topology model using an isolation forest algorithm to obtain optimized electric energy topology data; S3. Acquire environmental data, and compensate the optimized power topology data for environmental effects based on the environmental data to obtain compensated optimized power topology data; dynamically correct the clean power metering data based on the compensated optimized power topology data using a Kalman filter algorithm to obtain actual power metering data; and map the actual power metering data to standard power units using an edge computing node to obtain standardized power metering data; S4. Input the standardized electricity metering data into a pre-trained fusion model, which uses LSTM and CNN to extract time series features and fuse them with spatial features to obtain electricity usage pattern labels; compare the electricity usage pattern labels with preset electricity usage specification data, calculate the electricity usage compliance score, and obtain a smart metering report; wherein the smart metering report includes real-time metering values, electricity usage patterns, and compliance scores.

[0021] It should be noted that, in operation, a sensor array refers to a group of integrated sensor devices that work together to monitor and collect different types of data. In an electricity meter box, the sensor array may include temperature and humidity sensors, current sensors, power sensors, and other sensors to collect real-time energy and environmental data. 5G networks are the fifth generation of mobile communication networks. Compared to previous generations, 5G offers higher transmission speeds, lower latency, and greater network capacity. Its role in electricity metering systems is to quickly transmit raw energy data and environmental parameters, ensuring real-time and accurate data. Dynamic threshold filtering is a data processing technique that dynamically adjusts thresholds based on real-time data changes to remove noise or outliers. This improves data clarity, enabling subsequent analysis to be based on more accurate and clean data. Principal component analysis (PCA) is a commonly used dimensionality reduction technique that projects data from a high-dimensional space to a low-dimensional space through linear transformation to extract key features. In this method, PCA is used to extract the most important energy parameters from raw energy data, reducing data complexity. Graph neural networks are graph-based neural network algorithms capable of processing and analyzing graph data. In electricity metering, graph neural networks are used to build energy topology models. By analyzing the relationships between nodes in the power grid, they help identify the distribution and transmission of electricity. Isolation forest is a tree-based unsupervised learning algorithm primarily used for anomaly detection. It isolates data points by constructing multiple random trees. If a point is quickly isolated across multiple trees, it is considered an anomaly. In power topology, the isolation forest algorithm is used to detect abnormal nodes in the system, such as faulty equipment or erroneous power transmission. Electromagnetic interference (EMI) is noise or unwanted signals generated by electrical equipment, which can affect sensor measurement accuracy. EMI compensation models, combined with data such as ambient temperature and humidity, use mathematical models to compensate for errors caused by interference, ensuring accurate energy metering. Kalman filtering is an algorithm used for data estimation and correction. It leverages prior information and current observations to dynamically estimate system states. In energy metering, Kalman filtering can be used to correct for raw data and sensor noise, resulting in more accurate actual energy measurements. Edge computing is a distributed computing approach in which data processing is performed closer to the data source (such as a local server or device), reducing transmission latency and bandwidth requirements. In the energy metering system, edge computing nodes are responsible for real-time processing and analysis of energy data, thereby improving the system's responsiveness and real-time performance. LSTM (Long Short-Term Memory) is a specialized recurrent neural network (RNN) suitable for processing time series data. CNN (Convolutional Neural Network) excels at processing image and spatial data. In this approach, LSTM is used to extract time series features from energy data, while CNN is used to integrate spatial features. Combining these two methods enables efficient analysis of energy usage patterns.Energy usage pattern tags are generated by analyzing energy usage data to reflect energy usage behavior. For example, they can display device usage habits, load changes, or energy consumption trends. The electricity compliance score compares energy usage patterns with pre-set electricity usage specifications to assess whether energy usage complies with specified standards or energy conservation requirements. If electricity usage does not comply with the specifications, the score is lower, while if it does, the score is higher. Smart metering reports include real-time electricity metering data, electricity usage pattern analysis, and compliance scores. This report provides users or managers with detailed energy usage information, helping them make smarter decisions, optimize electricity usage, and reduce energy waste.

[0022] In one embodiment, a sensor array is deployed inside an electric energy meter box; raw electric energy data and environmental parameters are periodically collected via a 5G network to obtain a raw electric energy metering data stream, including: A modular sensor array architecture is designed inside the electricity meter box. The sensor array includes high-precision current transformers, voltage sensors, and temperature and humidity sensors. An ultra-low-latency communication link is established over the 5G network to periodically collect raw power data and environmental parameters, adding timestamps and device ID tags to obtain the raw power metering data stream. Perform dynamic threshold filtering on the raw data and use a wavelet transform denoising algorithm to remove harmonic interference. Use edge computing nodes to verify data integrity in real time to obtain clean electricity metering data. The clean data is input into the pre-trained graph neural network model to construct a loop power topology association diagram; features are extracted through the node embedding algorithm and combined with the LSTM time series prediction model to obtain the power consumption pattern label.

[0023] Dynamic wavelet Kalman filter formula: in, Indicated on scale and time delay The signal wavelet coefficients under , reflect the characteristics of the signal at different time positions and scales, Represents the raw energy metering data over time The changing signal represents the conjugate of the wavelet function, which represents the wavelet basis function at different scales. and time offset The following transformation, Represents the scale parameter, which controls the width of the wavelet basis function. The larger the scale, the lower the frequency component, and vice versa. Represents the denoised signal, the signal recovered after wavelet coefficient denoising, Indicates threshold-based The wavelet coefficients are processed to remove the noise. Usually, the parts with smaller absolute values ​​of the wavelet coefficients will be suppressed or discarded, and the larger coefficients will be retained to retain the useful components of the signal. represents the scaled wavelet function used to reconstruct the signal, represents the set of all scales, covering components in different frequency ranges, Represents the cleaned electric energy data, the final data obtained after wavelet denoising and Kalman filtering, represents the state transition matrix of the Kalman filter, which represents the dynamic change of the system from one time step to the next time step; represents the control input matrix of the Kalman filter, which represents the influence of the external control signal on the system state; Represents the control input signal, usually indicating the impact of external factors (such as environmental parameters, equipment status, etc.) on power usage.

[0024] In this energy metering system, the sensor array, designed with a modular architecture, includes high-precision current transformers, voltage sensors, and temperature and humidity sensors, enabling precise collection of energy and environmental parameters. A low-latency communication link is established over a 5G network to regularly collect raw energy data streams. Each data packet is timestamped and identified with a device ID to ensure data accuracy and traceability. The raw data undergoes dynamic threshold filtering and a wavelet transform denoising algorithm to remove grid harmonic interference and improve data quality. Edge computing nodes verify data integrity in real time to ensure data is not affected by network latency. The clean data is then fed into a pre-trained graph neural network model, which constructs a power topology correlation graph to analyze power flow patterns. A node embedding algorithm extracts grid topology features and, combined with an LSTM time series prediction model, predicts future power usage patterns and generates power usage pattern labels to guide users in optimizing their electricity usage and improving energy efficiency.

[0025] In one embodiment, dynamic threshold filtering is performed on the raw data to obtain clean electric energy metering data, including: A dynamic threshold benchmark library is built based on historical power data. The sliding window algorithm is used to calculate the mean and standard deviation of the data within the window to obtain the adaptive threshold range. An environmental interference model is established based on temperature and humidity sensor data to obtain the environmental adaptive threshold parameters. Perform real-time deviation detection on the raw power data stream and calculate the absolute error between the current data point and the sliding window mean. Use the soft threshold shrinkage method to denoise the high-frequency noise band to obtain preliminary denoised data. Cyclic redundancy check is used to detect the integrity of data packets to obtain verified intermediate data; The intermediate data is input into the improved Kalman filter, and the state estimation is performed by combining the dynamic threshold parameters and the wavelet denoising results; the data deviation is corrected through the prediction update cycle to obtain clean electricity metering data with a noise level below the threshold.

[0026] In this energy metering system, dynamic threshold filtering technology is used for real-time deviation detection. By establishing a dynamic threshold benchmark library and an environmental interference model, the data error threshold is adjusted in real time. A sliding window algorithm calculates the mean and standard deviation of the data, providing an adaptive threshold range for the raw energy data to accommodate the impact of environmental changes on energy metering. Environmental data from temperature and humidity sensors further optimizes the threshold parameters, making the system more resilient to environmental interference. Real-time deviation detection is combined with a wavelet denoising algorithm to remove high-frequency noise from the raw data. Data integrity is verified using a cyclic redundancy check to ensure data accuracy. Furthermore, an improved Kalman filter performs state estimation on the data, dynamically adjusting the threshold and denoising results to correct data deviations. Ultimately, clean energy metering data is output, ensuring data accuracy and stability. This process effectively reduces environmental interference and noise, ensuring high-quality energy metering data and providing accurate and reliable input data for subsequent energy analysis and forecasting.

[0027] In one embodiment, principal component analysis is performed on clean power metering data to identify key power parameters; a power topology model is constructed based on a graph neural network, and the power data of each loop is correlated to obtain initial power topology network data, including: The clean electricity metering data is Z-score standardized, and the principal components are screened by cumulative variance contribution rate to obtain a dimensionality-reduced feature matrix containing harmonic content, three-phase imbalance, and power factor. A multimodal graph structure is constructed, and each circuit in the energy meter box is defined as a graph node. The correlation between the energy parameters between nodes is calculated using the Pearson correlation coefficient to obtain the initial energy topology graph. Deploy a graph attention network and perform multi-layer graph convolution operations on the initial topology graph; dynamically adjust edge weights through the attention mechanism to obtain optimized power topology network data including node embeddings and edge weights.

[0028] Soft threshold shrinkage denoising formula: , in, Represents the denoised signal, which is the original signal The result after applying soft thresholding, represents the symbolic function, The symbol, represents the original signal, which may contain noise, Represents the soft threshold function, according to time Dynamically changing thresholds, It represents the standard deviation of the noise and describes the change of the noise intensity over time. It represents the noise component in the signal. Indicates the length or size of the data, Represents a parameter that controls the sensitivity of the threshold. It determines Flexibility over time, It is usually a local statistic of the signal, which may represent the variation or smoothness of the signal and is used to dynamically adjust the threshold.

[0029] This design first performs Z-score normalization on clean energy metering data to ensure data uniformity and comparability. Principal components are filtered using the cumulative variance contribution rate to extract key information from the energy data, such as harmonic content, three-phase imbalance, and power factor. This reduces the high-dimensional data into a feature matrix, reducing noise and redundant data. Next, a multimodal graph structure is constructed, with each circuit within the energy meter box as a node. The Pearson correlation coefficient is used to calculate the correlation between energy parameters at each node, thus forming an initial energy topology map. This topology map illustrates the energy flow relationships between circuits and provides a foundation for subsequent analysis. Finally, a graph attention network (GAT) is deployed to perform multi-layer graph convolution operations on the initial energy topology map, using an attention mechanism to dynamically adjust edge weights. Through this optimization approach, the energy topology network more accurately reflects the energy relationships between nodes (circuits), improving the relevance of energy data and the accuracy of predictions. This process facilitates a deeper understanding of energy flow patterns, providing data support for power optimization and fault detection.

[0030] In one embodiment, an isolation forest algorithm is used to detect abnormal nodes in a topological network and optimize the electric energy metering path to obtain optimized electric energy topology data, including: Extract node feature vectors from the initial power topology network data; construct a node feature matrix and retain edge connection relationships to obtain topological feature data suitable for isolation forest input; Deploy a multi-scale isolation forest model, randomly sample subspaces on the feature matrix, and construct multiple isolation trees. Calculate anomaly scores based on path length to obtain detection results that include abnormal node IDs and confidence levels. Optimize the network topology based on anomaly detection results, remove abnormal nodes or adjust their edge connection weights; combine the efficiency requirements of the energy metering path, and use the Dijkstra algorithm to recalculate the shortest path to obtain an optimized edge weight matrix and node list; The optimization effect is verified by simulating the injection of abnormal data, and the anomaly detection accuracy and path delay are calculated; when the performance does not meet the standards, the power topology network data with self-optimization capabilities is obtained.

[0031] In this energy metering system, the isolation forest algorithm is used to detect abnormal nodes in the energy topology network and optimize the energy metering path. First, node feature vectors are extracted from the initial topology network data, and a node feature matrix is ​​constructed, preserving edge connectivity relationships to meet the input requirements of the isolation forest. Then, a multi-scale isolation forest model is deployed to construct multiple isolation trees by randomly sampling subspaces. Path lengths are calculated to determine anomaly scores for each node, and node anomaly is determined based on these scores. After anomaly detection, the topology network is optimized to remove or adjust edge connection weights for abnormal nodes. Considering the efficiency of the energy metering path, the Dijkstra algorithm is used to recalculate the shortest path and optimize the edge weight matrix and node list in the topology network, thereby improving the efficiency of energy flow. Finally, the performance of the optimized topology network, including anomaly detection accuracy and path delay, is verified through simulations injecting abnormal data. If performance falls short of the target, the system will self-optimize to generate more efficient and adaptive energy topology network data. This optimization process helps ensure the reliability and accuracy of the energy metering system and enhances overall energy management effectiveness.

[0032] In one embodiment, environmental effect compensation is performed on the optimized power topology data, and an electromagnetic interference compensation model is established in combination with temperature and humidity sensor data. The metering data is dynamically corrected using a Kalman filter algorithm to obtain the actual power metering value after compensation: Extract node features and topological attributes from the optimized power topology data; synchronously collect temperature and humidity sensor data to construct a multimodal dataset containing power parameters and environmental parameters to obtain the environmental power correlation feature matrix; An electromagnetic interference compensation model is constructed based on a graph attention network. The nonlinear effect of the environment on power parameters is learned through multi-layer graph convolution to obtain the node-level electromagnetic interference compensation coefficient. The compensation coefficient is applied to optimize the power topology data, and reverse corrections are performed on parameters such as voltage and current to obtain preliminary corrected data after compensation. Edge computing nodes are deployed to perform Kalman filtering, using the compensation data as observations and the model prediction values ​​as state estimates. Noise is dynamically corrected through a prediction update loop to obtain actual power metering values ​​with noise levels below a threshold. The compensation effect is verified by simulating environmental change experiments, and the data errors before and after correction are calculated to obtain the actual electric energy metering value with environmental adaptability.

[0033] The design first extracts node features and topological attributes from the optimized power topology data, while simultaneously collecting temperature and humidity sensor data to construct a multimodal dataset containing power parameters and environmental parameters, thereby generating an environmental power correlation feature matrix. Next, an electromagnetic interference compensation model is constructed using a graph attention network (GAT). Through multi-layer graph convolution, the nonlinear effects of the environment on power parameters are learned to determine the electromagnetic interference compensation coefficient for each node. This compensation coefficient is applied to the optimized power topology data, correcting parameters such as voltage and current to generate preliminary compensation data. Edge computing nodes are then deployed, and the Kalman filter algorithm dynamically corrects the compensated data. Using the compensated data as observations, the Kalman filter updates the state estimate with predicted values, dynamically correcting for noise and ensuring that the final power metering noise level remains below a set threshold. Finally, the compensation effect is verified through experiments simulating environmental changes, calculating the data errors before and after correction to ensure the system's robust environmental adaptability and high-precision actual power metering.

[0034] In one embodiment, edge computing nodes are used to map actual metering values ​​to standard electrical energy units to obtain standardized electrical energy metering data, including: Deploy a unit conversion engine at the edge computing node to receive the actual compensated energy metering value. Use a pre-trained neural network model to identify the energy parameter type to obtain initial standardized data. Deploy a dynamic calibration strategy, combine the historical data of standard electricity meters stored in edge nodes, calculate the deviation between the initial standardized data and the standard value, and obtain the calibrated intermediate standardized data; The LSTM time series model within the edge computing node is used to perform nonlinear error correction on intermediate standardized data. By capturing the temporal correlation of power parameters, the accumulated error during unit conversion is predicted and corrected to obtain high-precision standardized power metering data. The standardized data is synchronized to the cloud blockchain platform, and the data fingerprint is obtained through the hash algorithm and stored on the chain to obtain a standardized electricity metering data stream with tamper-proof characteristics.

[0035] In this design, the node first receives compensated energy metering data and uses a pretrained neural network model to identify energy parameter types, generating initial standardized data. Subsequently, a dynamic calibration strategy is deployed, combining historical standard energy meter data to calculate the deviation between the initial standardized data and the standard value, thereby obtaining intermediate standardized data. Next, an LSTM time series model is used to perform nonlinear error correction on the intermediate data, capturing the temporal correlation of energy parameters and predicting and correcting accumulated errors during unit conversion. Ultimately, high-precision standardized energy metering data is obtained. This standardized data is synchronized to a cloud-based blockchain platform, where a data fingerprint is generated using a hash algorithm and uploaded to the blockchain, ensuring that the data cannot be tampered with, thereby enhancing its credibility and security. This process not only improves the accuracy of energy metering but also ensures the tamper-proof nature of the data stream, providing reliable support for power management and data monitoring.

[0036] In one embodiment, the standardized electric energy metering data is input into a pre-trained LSTM-CNN fusion model, and the electric energy usage pattern labels are obtained by extracting temporal features and fusing them with spatial features, including: Standardized electricity metering data is divided into multi-scale time windows to obtain short-term, medium-term, and long-term time series segments. The frequency domain features of each segment are extracted through wavelet transform, and a multimodal feature matrix is ​​constructed by combining time domain statistics to obtain enhanced data suitable for LSTM-CNN input. A parallel LSTM-CNN architecture is deployed. The LSTM branch captures the dynamic changes in time series segments, while the CNN branch extracts the spatial distribution patterns of frequency domain features. The outputs of these two branches are fused through a cross-attention mechanism to generate a fused feature vector that reflects temporal and spatial correlations. This design begins by partitioning the normalized data into multi-scale time windows, forming short, medium, and long time series segments. Next, frequency domain features are extracted through a wavelet transform and combined with time domain statistics to construct a multimodal feature matrix, enhancing the data's expressiveness and making it suitable for LSTM-CNN input. The model architecture utilizes parallel LSTM and CNN branches. The LSTM branch captures the dynamic changes in time series segments, while the CNN branch focuses on extracting the spatial distribution patterns of frequency domain features. Through the cross-attention mechanism, the outputs of the LSTM and CNN are effectively fused to generate a fused feature vector that contains both temporal and spatial correlation information. This vector more accurately describes energy usage patterns and supports subsequent intelligent analysis, anomaly detection, and prediction tasks.

[0037] In one embodiment, the standardized electric energy metering data is input into a pre-trained LSTM-CNN fusion model, and the electric energy usage pattern label is obtained by extracting temporal features and fusing spatial features, further comprising: The fused feature vector is input into the pre-trained classification layer and combined with the power usage scenario label library for pattern matching. The classification threshold is dynamically adjusted using a reinforcement learning algorithm to obtain the power usage pattern label. The accuracy of labels is verified by back-testing historical power data, and the classification F1 value is calculated to obtain a label flow of power usage patterns with adaptive capabilities.

[0038] This design feeds standardized electricity metering data into the LSTM-CNN model, extracts the spatial distribution patterns of time series dynamics and frequency domain features, and generates a fused feature vector. Then, combined with a library of electricity usage scenario labels, the fused feature vector is fed into a pre-trained classification layer for pattern matching to identify electricity usage patterns. To improve classification accuracy, the system uses a reinforcement learning algorithm to dynamically adjust the classification threshold and optimize the label matching process. This strategy adaptively adjusts the model's response to different usage scenarios, enhancing the system's adaptability to complex patterns. Finally, the accuracy of the pattern labels is verified by reviewing historical electricity data, and the classification F1 value is calculated to evaluate the model's performance. This process ensures that the generated electricity usage pattern labels are adaptive and can accurately reflect different electricity usage scenarios, providing precise data support for power management and forecasting.

[0039] The real-time metering system of the intelligent modular electric energy metering box is applicable to the real-time metering method of the intelligent modular electric energy metering box, including: The data filtering unit 1 is used to deploy a sensor array inside the electric energy metering box; collect raw electric energy data and environmental parameters through the sensor array to obtain a raw electric energy metering data stream; and perform dynamic threshold filtering on the raw electric energy metering data stream to obtain clean electric energy metering data; Power topology unit 2 is used to perform principal component analysis on clean power metering data to obtain key power parameters; construct a power topology model based on a graph neural network, input the key power parameters into the power topology model to obtain initial power topology network data; and detect abnormal nodes in the power topology model using an isolation forest algorithm to obtain optimized power topology data. The data correction unit 3 is used to obtain environmental data, and compensate the optimized power topology data for environmental effects based on the environmental data to obtain compensated optimized power topology data; dynamically correct the clean power metering data based on the compensated optimized power topology data using a Kalman filter algorithm to obtain actual power metering data; and use edge computing nodes to map the actual power metering data to standard power units to obtain standardized power metering data; Electricity usage specification unit 4 is used to input standardized electricity metering data into a pre-trained LSTM-CNN fusion model. By extracting time series features and fusing spatial features, it obtains electricity usage pattern labels. The electricity usage pattern labels are compared with the preset electricity usage specification data, and the electricity compliance score is calculated to obtain a smart metering report. Among them, the smart metering report includes real-time metering values, electricity usage patterns and compliance scores.

[0040] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A real-time metering method for an intelligent modular electric energy metering box, characterized in that: include: Deploy a sensor array inside the electricity meter box; Collecting raw power data and environmental parameters through the sensor array to obtain a raw power metering data stream; Performing dynamic threshold filtering on the raw electric energy metering data stream to obtain clean electric energy metering data; Performing principal component analysis on the clean electric energy metering data to obtain key electric energy parameters; Constructing an electric energy topology model based on a graph neural network, inputting the key electric energy parameters into the electric energy topology model to obtain initial electric energy topology network data; detecting abnormal nodes in the electric energy topology model using an isolation forest algorithm to obtain optimized electric energy topology data; Acquiring environmental data, and performing environmental effect compensation on the optimized power topology data according to the environmental data to obtain compensated optimized power topology data; Dynamically correcting the clean electric energy metering data according to the compensated optimized electric energy topology data through a Kalman filter algorithm to obtain actual electric energy metering data; Mapping the actual electric energy metering data to standard electric energy units using edge computing nodes to obtain standardized electric energy metering data; The standardized electricity metering data is input into a pre-trained fusion model. The fusion model uses LSTM and CNN to extract time series features and fuse them with spatial features to obtain electricity usage pattern labels. The electricity usage pattern labels are compared with preset electricity usage specification data to calculate the electricity usage compliance score to obtain a smart metering report. The smart metering report includes real-time metering values, electricity usage patterns, and compliance scores.

2. The real-time metering method of the intelligent modular electric energy meter box according to claim 1 is characterized in that: Deploy a sensor array inside the electricity meter box; The sensor array collects raw power data and environmental parameters to obtain a raw power metering data stream, including: Design a modular sensor array architecture inside the electric energy meter box, the sensor array including high-precision current transformers, voltage sensors, and temperature and humidity sensors; An ultra-low latency communication link is established through the network, raw power data and environmental parameters are periodically collected, and timestamps and device ID tags are added to obtain the raw power metering data stream.

3. The real-time metering method of the intelligent modular electric energy meter box according to claim 2, characterized in that: Perform dynamic threshold filtering on the raw data to obtain clean energy metering data, including: A dynamic threshold benchmark library is built based on historical power data. The sliding window algorithm is used to calculate the mean and standard deviation of the data within the window to obtain the adaptive threshold range. An environmental interference model is established based on temperature and humidity sensor data to obtain the environmental adaptive threshold parameters. Perform real-time deviation detection on the raw power data stream and calculate the absolute error between the current data point and the sliding window mean. Use the soft threshold shrinkage method to denoise the high-frequency noise band to obtain preliminary denoised data. Cyclic redundancy check is used to detect the integrity of the data packet to obtain the verified intermediate data; The verified intermediate data is input into the improved Kalman filter, and state estimation is performed in combination with the dynamic threshold parameter and the wavelet denoising result; the data deviation is corrected through the prediction update cycle to obtain clean electricity metering data.

4. The real-time metering method of the intelligent modular electric energy metering box according to claim 3 is characterized in that: Performing principal component analysis on the clean electricity metering data to identify key electricity parameters; Build an electric energy topology model based on the graph neural network and associate the electric energy data of each circuit to obtain the initial electric energy topology network data, including: The clean electric energy metering data is standardized, and the principal components are screened by cumulative variance contribution rate to obtain a reduced dimension feature matrix; the reduced dimension feature matrix includes harmonic content, three-phase imbalance and power factor; A multimodal graph structure is constructed, and each circuit in the electric energy meter box is defined as a graph node. The correlation between the electric energy parameters of the graph nodes is calculated using the Pearson correlation coefficient to obtain an initial electric energy topology graph. Deploy a graph attention network and perform multi-layer graph convolution operations on the initial topology graph; dynamically adjust edge weights through the attention mechanism to obtain optimized power topology network data.

5. The real-time metering method of the intelligent modular electric energy meter box according to claim 4 is characterized in that: The isolation forest algorithm is used to detect abnormal nodes in the topology network and optimize the energy metering path to obtain optimized energy topology data, including: Extract node feature vectors from the initial power topology network data; construct a node feature matrix and retain edge connection relationships to obtain topological feature data suitable for isolation forest input; An isolation forest model is constructed, and a plurality of isolation trees are constructed by randomly sampling subspaces on the node feature matrix; an anomaly score is calculated by path length to obtain an anomaly detection result; Optimizing the topology network based on the anomaly detection results and removing abnormal nodes; calculating the shortest path of the topology network optimized by the anomaly detection results in combination with the efficiency requirements of the electric energy metering path to obtain an optimized edge weight matrix and node list; The optimization effect is verified by simulating the injection of abnormal data, and the anomaly detection accuracy and path delay are calculated to obtain the optimized power topology network data.

6. The real-time metering method of the intelligent modular electric energy meter box according to claim 5, characterized in that: The optimized power topology data is compensated for environmental effects, and an electromagnetic interference compensation model is established in combination with the temperature and humidity sensor data. The metering data is dynamically corrected using the Kalman filter algorithm to obtain the actual power metering value after compensation: extracting node features and topological attributes from the optimized power topology data; Collect temperature and humidity sensor data and construct a multimodal dataset containing power parameters and environmental parameters to obtain the environmental power correlation feature matrix; An electromagnetic interference compensation model is constructed based on a graph attention network, and the nonlinear effect of the environment on the multimodal dataset is learned through multi-layer graph convolution to obtain node-level electromagnetic interference compensation coefficients; Applying the node-level electromagnetic interference compensation coefficient to the optimized power topology data, and performing reverse correction on parameters such as voltage and current to obtain preliminary corrected data after compensation; Deploy edge computing nodes to perform Kalman filtering, using compensation data as observations and model predictions as state estimates. Dynamically correct noise through a prediction update loop to obtain actual energy metering values ​​with noise levels below a threshold. The compensation effect is verified by simulating environmental change experiments, and the data errors of the actual electric energy measurement values ​​before and after correction are calculated to obtain the actual electric energy measurement values ​​after compensation.

7. The real-time metering method of the intelligent modular electric energy meter box according to claim 6, characterized in that: Use edge computing nodes to map actual metering values ​​to standard energy units to obtain standardized energy metering data, including: Deploy a unit conversion engine at the edge computing node to receive the actual compensated energy metering value. Use a pre-trained neural network model to identify the energy parameter type to obtain initial standardized data. Deploy a dynamic calibration strategy, combine the historical data of the standard electric energy meter stored in the edge node, calculate the deviation between the initial standardized data and the standard value, and obtain intermediate standardized data; A timing model within the edge computing node is used to perform nonlinear error correction on the intermediate standardized data; by capturing the time correlation of the electric energy parameters, the accumulated error in the unit conversion process is predicted and corrected to obtain standardized electric energy metering data; The standardized electricity metering data is synchronized to the cloud blockchain platform, and a data fingerprint is generated through a hash algorithm and stored on the chain to obtain a standardized electricity metering data stream.

8. The real-time metering method of the intelligent modular electric energy metering box according to claim 7, characterized in that: The standardized electric energy metering data is input into the pre-trained fusion model, and the electric energy usage pattern label is obtained by extracting the temporal features and fusing them with the spatial features, including: The standardized electric energy metering data is divided into multi-scale time windows to obtain short-time, medium-time, and long-time time series segments; the frequency domain features of each segment are extracted by wavelet transform, and a multimodal feature matrix is ​​constructed by combining time domain statistics to obtain enhanced data; A parallel LSTM and CNN architecture is deployed. The LSTM branch captures the dynamic changes of the enhanced data, and the CNN branch extracts the spatial distribution pattern of frequency domain features. The outputs of the two branches are fused through a cross-attention mechanism to obtain a fused feature vector.

9. The real-time metering method of the intelligent modular electric energy meter box according to claim 8, characterized in that: The standardized electric energy metering data is input into a pre-trained fusion model, and a label of electric energy usage pattern is obtained by extracting temporal features and fusing spatial features, further comprising: Inputting the fused feature vector into a pre-trained classification layer and combining it with a power usage scenario label library for pattern matching; dynamically adjusting the classification threshold using a reinforcement learning algorithm to obtain a power usage pattern label; The accuracy of labels is verified by back-tracing historical power data, and the classification F1 value is calculated to obtain the power usage pattern label flow.

10. A real-time metering system for an intelligent modular electric energy metering box, which is applicable to the real-time metering method for an intelligent modular electric energy metering box according to any one of claims 1 to 9, characterized in that: include: A data filtering unit (1) is provided, wherein a sensor array is deployed inside the electric energy metering box; Collecting raw power data and environmental parameters through the sensor array to obtain a raw power metering data stream; Performing dynamic threshold filtering on the raw electric energy metering data stream to obtain clean electric energy metering data; An electric energy topology unit (2) performs principal component analysis on the clean electric energy metering data to obtain key electric energy parameters; Constructing an electric energy topology model based on a graph neural network, inputting the key electric energy parameters into the electric energy topology model to obtain initial electric energy topology network data; detecting abnormal nodes in the electric energy topology model using an isolation forest algorithm to obtain optimized electric energy topology data; A data correction unit (3) acquires environmental data and performs environmental effect compensation on the optimized power topology data according to the environmental data to obtain compensated optimized power topology data; Dynamically correcting the clean electric energy metering data according to the compensated optimized electric energy topology data through a Kalman filter algorithm to obtain actual electric energy metering data; Mapping the actual electric energy metering data to standard electric energy units using edge computing nodes to obtain standardized electric energy metering data; The electricity usage specification unit (4) inputs the standardized electricity metering data into a pre-trained fusion model, extracts time series features and fuses spatial features to obtain an electricity usage pattern label; compares the electricity usage pattern label with preset electricity usage specification data, calculates an electricity usage compliance score, and obtains an intelligent metering report; wherein the intelligent metering report includes a real-time metering value, an electricity usage pattern, and a compliance score.

Images