Intelligent electric energy meter based on artificial intelligence and metering regulation and control method

Abstract

The invention discloses an intelligent electric energy meter based on artificial intelligence and a metering regulation and control method. The method comprises the steps that an acquisition module acquires metering deviation data and multi-dimensional power consumption data; the correction module is used for correcting the multi-dimensional power consumption data according to a pre-constructed dynamic compensation model and the metering deviation data to obtain accurate power consumption data; an analysis module performs data analysis on the accurate power consumption data to obtain a target power consumption data set; the behavior extraction module is used for extracting differentiated power utilization characteristics and dynamic power utilization modes of users from the target power utilization data set; and the generation module generates a metering self-adaptive regulation and control scheme according to the differentiated power utilization characteristics and the dynamic power utilization mode, and executes the metering self-adaptive regulation and control scheme. The data processing precision and the analysis depth are improved, and the metering precision is improved by combining a dynamic compensation model with real-time metering deviation data; and the metering self-adaptive regulation and control scheme is generated by extracting the differentiated electricity utilization characteristics and the dynamic electricity utilization modes of the users, so that the metering accuracy of the intelligent electric energy meter is improved.

Description

Technical Field

[0001] This invention relates to the field of electricity meter technology, and in particular to an intelligent electricity meter based on artificial intelligence and a metering and control method. Background Technology

[0002] With the rapid advancement of technology, electricity has become an extremely important energy source for social development. Against this backdrop, continuously improving the reliability of power grid operation using advanced technologies is a core objective pursued by the power industry. In particular, smart meters, as one of the key technologies in the process of power grid intelligentization, have greatly promoted the effectiveness of power resource management and have been widely applied in various sectors of society. Smart meters can accurately measure users' electricity consumption and increase the transparency of power companies' information, thereby giving users greater access to information.

[0003] However, in practical applications, electricity meters need improvement in metering accuracy and automation. They still have significant shortcomings in complex electricity usage scenarios. Relying on preset fixed compensation parameters, they cannot adapt in real time to metering deviations caused by dynamic factors such as grid fluctuations (e.g., harmonic interference), component aging, and sudden changes in user load. Metering accuracy still needs improvement in scenarios such as high industrial loads and simultaneous operation of multiple household devices. They do not analyze the differentiated electricity usage characteristics of users (e.g., the differences in electricity usage patterns between residential and commercial users), and the control strategy is a uniform preset mode (e.g., fixed threshold protection). They cannot optimize metering methods or provide personalized electricity usage suggestions based on users' dynamic electricity usage habits, which can easily lead to metering errors or waste of resources. In the process of data analysis, the data processing accuracy and analysis depth are limited, and accurate electricity usage data cannot be obtained. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the aforementioned technologies. Therefore, the purpose of this invention is to propose an artificial intelligence-based smart energy meter and metering control method, which improves data processing accuracy and analysis depth, obtains accurate electricity consumption data, improves metering accuracy by combining a dynamic compensation model with real-time metering deviation data, and generates an adaptive metering control scheme by extracting user-differentiated electricity consumption characteristics and dynamic electricity consumption patterns, thereby facilitating improved metering accuracy of the smart energy meter.

[0005] To achieve the above objectives, embodiments of the present invention propose an artificial intelligence-based smart energy meter, comprising:

[0006] The acquisition module is used to acquire metering deviation data and multi-dimensional electricity consumption data;

[0007] The correction module is used to correct multi-dimensional electricity consumption data based on a pre-built dynamic compensation model and metering deviation data to obtain accurate electricity consumption data.

[0008] The analysis module is used to perform data analysis on precise electricity consumption data to obtain the target electricity consumption dataset;

[0009] The behavior extraction module is used to extract user-differentiated electricity consumption characteristics and dynamic electricity consumption patterns from the target electricity consumption dataset;

[0010] The generation module is used to generate metering adaptive control schemes based on differentiated electricity consumption characteristics and dynamic electricity consumption patterns, and to execute the metering adaptive control schemes.

[0011] According to some embodiments of the present invention, the acquisition module is based on the power line carrier communication module and the LoRa communication module to connect to the data monitoring module, and acquires metering deviation data and multi-dimensional electricity consumption data based on a sliding window.

[0012] According to some embodiments of the present invention, a method for determining a dynamic compensation model by a correction module includes:

[0013] Obtain historical measurement deviation data and corresponding target compensation values;

[0014] Feature mining is performed on historical measurement deviation data to extract key influencing factors and error trends over a recent preset time period; key influencing factors include environmental parameters, power grid parameters, equipment parameters, and usage parameters;

[0015] An initial compensation model is constructed. The initial compensation model includes a feedforward neural network and a timing compensator based on a GRU network. The feedforward neural network includes an input layer, two hidden layers, and an output layer. The number of nodes in the two hidden layers are 32 and 16, respectively.

[0016] Key influencing factors are input into a feedforward neural network, which outputs a basic error compensation value; errors within a near-preset time period are input into a time series compensator, which outputs a dynamic correction value; the two are weighted and summed to obtain the predicted compensation value.

[0017] The predicted compensation value is compared with the target compensation value, and the initial compensation model is iteratively trained based on the comparison results to obtain a trained dynamic compensation model.

[0018] According to some embodiments of the present invention, the correction module includes:

[0019] The first determining module is used to input the measurement deviation data into the pre-built dynamic compensation model to determine the dynamic compensation value;

[0020] The compensation module is used to correct multi-dimensional electricity consumption data based on dynamic compensation values ​​to obtain accurate electricity consumption data.

[0021] According to some embodiments of the present invention, the analysis module includes:

[0022] The clustering module is used to perform pattern clustering on precise electricity consumption data to obtain clustering results;

[0023] The assessment module is used to perform risk assessments on precise electricity consumption data and obtain assessment results.

[0024] The screening module is used to screen for anomalies based on clustering and evaluation results, remove abnormal data, and obtain intermediate data. The intermediate data is smoothed by Gaussian filtering, and missing time period data is supplemented by interpolation based on the trend of adjacent time periods to obtain the target electricity consumption dataset.

[0025] According to some embodiments of the present invention, a clustering module includes:

[0026] The segmentation module is used to divide the precise electricity consumption data into several time-series segments and extract the feature vector of each time-series segment.

[0027] The partitioning module is used to initially partition clusters based on feature vectors using the density clustering DBSCAN method, and then perform temporal alignment of the clusters based on the dynamic time warping algorithm.

[0028] The second determining module is used for:

[0029] After time alignment, boundary data points are determined; wherein, boundary data points are data points whose distance difference to the center of two clusters is less than a first preset threshold and whose local density is less than a second preset threshold of the average density within the cluster.

[0030] The feature similarity between the boundary data points and each cluster is calculated, and the affiliation information of the boundary data points is determined by the fuzzy membership function to obtain the clustering results.

[0031] According to some embodiments of the present invention, the evaluation module includes:

[0032] The module is used to construct a multi-layered risk assessment system. The bottom layer of the multi-layered risk assessment system consists of data risk factors, the middle layer consists of equipment risk factors, and the top layer consists of user risk factors. Data risk factors include the frequency of metering value jumps and the number of times parameters exceed limits. Equipment risk factors include the duration of chip temperature exceeding the standard and the response delay of communication modules. User risk factors include the load mutation rate, the duration of abnormal power consumption at night, and the three-phase imbalance.

[0033] The first calculation module is used to input accurate electricity consumption data into the multi-layer risk assessment system and use failure mode and effects analysis to calculate the failure score of risk factors.

[0034] The second calculation module is used to calculate the risk factor weights using the information entropy method.

[0035] The third calculation module performs a weighted summation based on the failure scores and weights of the risk factors to obtain the risk value; based on the risk value and a preset risk level data table, it obtains the risk level as the evaluation result.

[0036] According to some embodiments of the present invention, a behavior extraction module includes:

[0037] The third determining module is used for:

[0038] Based on the time-series feature decomposition method, the target electricity consumption dataset is decomposed into trend terms, periodic terms, and random terms; long-term features are extracted from the trend terms; and short-term features are extracted from the periodic terms.

[0039] Based on long-term and short-term characteristics, determine the user's initial electricity consumption characteristics;

[0040] User profiles are constructed based on users' initial electricity consumption characteristics. These user profiles are then compared with a profile database, and the user's differentiated electricity consumption characteristics are determined based on the comparison results.

[0041] The mining module is used to perform pattern mining on the target electricity consumption dataset to identify dynamic electricity consumption patterns; the dynamic electricity consumption patterns include burst patterns, periodic patterns and gradual patterns.

[0042] According to some embodiments of the present invention, the generation module includes:

[0043] The control module is used to control target stratification based on differentiated electricity consumption characteristics;

[0044] The matching module is used to match differentiated control strategies based on dynamic power consumption patterns;

[0045] The fourth determination module is used to determine the adaptive metering control scheme based on the target stratification and differentiated control strategy.

[0046] According to some embodiments of the present invention, the metering and control method for an artificial intelligence-based smart energy meter as described above includes:

[0047] Acquire metering deviation data and multi-dimensional electricity consumption data;

[0048] Based on the pre-built dynamic compensation model and metering deviation data, multi-dimensional electricity consumption data is corrected to obtain accurate electricity consumption data;

[0049] Data analysis is performed on precise electricity consumption data to obtain the target electricity consumption dataset;

[0050] Extract user-differentiated electricity consumption characteristics and dynamic electricity consumption patterns from the target electricity consumption dataset;

[0051] A metering adaptive control scheme is generated based on differentiated electricity consumption characteristics and dynamic electricity consumption patterns, and then the metering adaptive control scheme is executed.

[0052] This invention proposes an artificial intelligence-based smart energy meter and metering control method. Through a dynamic compensation model, it performs real-time correction of multi-dimensional electricity consumption data, adapting to metering deviations caused by dynamic factors such as grid fluctuations, environmental changes, and component aging. This solves the problem that traditional static compensation mechanisms cannot handle errors in complex scenarios, improving metering accuracy and ensuring the accuracy and reliability of electricity consumption data. By extracting differentiated user electricity consumption characteristics and dynamic electricity consumption patterns, the generated adaptive control scheme can match the electricity consumption habits of different users, avoiding the insufficient adaptability of traditional unified control strategies and facilitating improved metering accuracy of smart energy meters.

[0053] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0054] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0055] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0056] Figure 1 This is a block diagram of an artificial intelligence-based smart energy meter according to an embodiment of the present invention;

[0057] Figure 2 This is a block diagram of an analysis module according to an embodiment of the present invention;

[0058] Figure 3 This is a flowchart of a metering and control method for an artificial intelligence-based smart energy meter according to an embodiment of the present invention. Detailed Implementation

[0059] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0060] like Figure 1 As shown in the figure, an embodiment of the present invention proposes an artificial intelligence-based smart energy meter, comprising:

[0061] The acquisition module is used to acquire metering deviation data and multi-dimensional electricity consumption data;

[0062] The correction module is used to correct multi-dimensional electricity consumption data based on a pre-built dynamic compensation model and metering deviation data to obtain accurate electricity consumption data.

[0063] The analysis module is used to perform data analysis on precise electricity consumption data to obtain the target electricity consumption dataset;

[0064] The behavior extraction module is used to extract user-differentiated electricity consumption characteristics and dynamic electricity consumption patterns from the target electricity consumption dataset;

[0065] The generation module is used to generate metering adaptive control schemes based on differentiated electricity consumption characteristics and dynamic electricity consumption patterns, and to execute the metering adaptive control schemes.

[0066] The working principle of the above technical solution is as follows: Metering deviation data consists of parameters related to metering errors caused by power grid fluctuations, environmental interference, and component aging; multi-dimensional electricity consumption data includes voltage, current, power, electricity consumption time period, load changes, and other multi-dimensional information. The metering deviation data is input into a dynamic compensation model to determine dynamic compensation values. Based on these dynamic compensation values, the multi-dimensional electricity consumption data is corrected to eliminate errors caused by various dynamic factors, outputting accurate electricity consumption data and solving the problem of insufficient accuracy in traditional static compensation. The analysis module performs data analysis on the accurate electricity consumption data, including pattern clustering, risk assessment, anomaly screening, smoothing, and data supplementation, facilitating the filtering of invalid information and retaining key data, thereby improving the accuracy of the target electricity consumption dataset. The behavior extraction module extracts differentiated electricity consumption characteristics and dynamic electricity consumption patterns from the target electricity consumption dataset. These differentiated characteristics include electricity consumption growth rate, seasonal fluctuation coefficient, load rate distribution range, peak hour start and end points, peak duration, peak-valley load ratio, and electricity consumption percentage for each time period. Based on these differentiated characteristics, the module determines the user's electricity consumption type, such as residential, commercial, and industrial. Dynamic electricity consumption patterns include sudden, periodic, and gradual patterns. A multi-dimensional control matrix is ​​generated based on preset user electricity consumption types, dynamic electricity consumption patterns, and corresponding metering and control schemes. Each user electricity consumption type and dynamic electricity consumption pattern corresponds to a metering and control scheme. Finally, an adaptive metering control scheme is generated based on the multi-dimensional control matrix according to the differentiated electricity consumption characteristics and dynamic electricity consumption patterns, and this adaptive metering control scheme is executed.

[0067] The beneficial effects of the above technical solution are as follows: By using a dynamic compensation model to correct multi-dimensional electricity consumption data in real time, it can adapt to metering deviations caused by dynamic factors such as power grid fluctuations, environmental changes, and component aging. This solves the problem that traditional static compensation mechanisms cannot cope with errors in complex scenarios, improving metering accuracy and ensuring the accuracy and reliability of electricity consumption data. Furthermore, by extracting differentiated user electricity consumption characteristics and dynamic electricity consumption patterns, the generated adaptive control scheme can match the electricity consumption habits of different users, avoiding the insufficient adaptability of traditional unified control strategies and facilitating improvements in the metering accuracy of smart meters.

[0068] According to some embodiments of the present invention, the acquisition module is based on the power line carrier communication module and the LoRa communication module to connect to the data monitoring module, and acquires metering deviation data and multi-dimensional electricity consumption data based on a sliding window.

[0069] The working principle and beneficial effects of the above technical solution are as follows: The acquisition module achieves efficient acquisition of metering deviation data and multi-dimensional electricity consumption data through the collaborative approach of dual communication modules and sliding window data acquisition. This solves the problems of single communication methods being susceptible to interference and having limited transmission distance in complex power grid environments. The sliding window mechanism achieves rolling data acquisition by dynamically capturing time segments, thereby improving the efficiency and accuracy of data acquisition.

[0070] According to some embodiments of the present invention, a method for determining a dynamic compensation model by a correction module includes:

[0071] Obtain historical measurement deviation data and corresponding target compensation values;

[0072] Feature mining is performed on historical measurement deviation data to extract key influencing factors and error trends over a recent preset time period; key influencing factors include environmental parameters, power grid parameters, equipment parameters, and usage parameters;

[0073] An initial compensation model is constructed. The initial compensation model includes a feedforward neural network and a timing compensator based on a GRU network. The feedforward neural network includes an input layer, two hidden layers, and an output layer. The number of nodes in the two hidden layers are 32 and 16, respectively.

[0074] Key influencing factors are input into a feedforward neural network, which outputs a basic error compensation value; errors within a near-preset time period are input into a time series compensator, which outputs a dynamic correction value; the two are weighted and summed to obtain the predicted compensation value.

[0075] The predicted compensation value is compared with the target compensation value, and the initial compensation model is iteratively trained based on the comparison results to obtain a trained dynamic compensation model.

[0076] The working principle of the above technical solution is as follows: Historical metering deviation data and corresponding target compensation values ​​are acquired; feature mining is performed on the historical metering deviation data to extract key influencing factors and error trends over a recent preset time period; environmental parameters include temperature and humidity; power grid parameters include voltage fluctuations, frequency deviations, and harmonic content; equipment parameters include chip aging and sampling circuit drift; usage parameters include user load levels and electricity usage periods; the error trend over a recent preset time period represents the slope and frequency of error changes over that period, used to capture the temporal evolution of the error. An initial compensation model is constructed, consisting of a feedforward neural network and a GRU network time-series compensator. The input layer of the feedforward neural network receives key influencing factors, and two hidden layers (with 32 and 16 nodes respectively, extracting complex mapping relationships between features through nonlinear activation functions) output a basic error compensation value based on the output layer, used to reflect the systematic error compensation caused by static factors. The error over a recent preset time period is input into the time-series compensator, which outputs a dynamic correction amount, reflecting the dynamic change compensation of the error over time. The two are weighted and summed to obtain the predicted compensation value, taking into account both static influences and dynamic changes. The predicted compensation value is compared with the target compensation value. Based on the comparison result, the deviation is quantified using the loss function. The weights of the feedforward neural network, the gating parameters of the GRU network, and the weighting coefficients are iteratively optimized using the backpropagation algorithm until the model prediction accuracy meets the preset threshold. Finally, the trained dynamic compensation model is obtained.

[0077] The beneficial effects of the above technical solution are as follows: By extracting multi-dimensional key influencing factors and combining them with the GRU network to capture time-series trends, it covers both static systematic errors (such as sampling drift caused by ambient temperature) and dynamic random errors (such as instantaneous deviations caused by sudden fluctuations in the power grid). The two hidden layers of the feedforward neural network effectively mine the interaction between key influencing factors through deep nonlinear mapping; the GRU network solves the gradient vanishing problem of traditional RNNs, accurately captures the long-term and short-term dependencies of error trends (such as the slow accumulation of errors over time caused by equipment aging), facilitates the output of predicted compensation values, meets real-time compensation requirements, and obtains an accurate dynamic compensation model.

[0078] According to some embodiments of the present invention, the correction module includes:

[0079] The first determining module is used to input the measurement deviation data into the pre-built dynamic compensation model to determine the dynamic compensation value;

[0080] The compensation module is used to correct multi-dimensional electricity consumption data based on dynamic compensation values ​​to obtain accurate electricity consumption data.

[0081] The working principle and beneficial effects of the above technical solution are as follows: The first determining module inputs the real-time collected metering deviation data into a pre-trained dynamic compensation model. Based on the input metering deviation data, combined with key influencing factors (such as environment, power grid, and equipment parameters) and recent error trends, the dynamic compensation model outputs corresponding dynamic compensation values, namely, basic compensation for static systematic errors and real-time correction for dynamic instantaneous errors. The compensation module then performs targeted corrections for multi-dimensional electricity consumption data based on the dynamic compensation values. Specifically, the dynamic compensation values ​​are decomposed according to the error characteristics of the data dimensions; for example, compensation coefficients for voltage deviations and correction amounts for power fluctuations are obtained. The original collected data is adjusted using a reverse calibration algorithm to eliminate the influence of metering deviations, facilitating accurate and precise electricity consumption data.

[0082] like Figure 2 As shown, according to some embodiments of the present invention, the analysis module includes:

[0083] The clustering module is used to perform pattern clustering on precise electricity consumption data to obtain clustering results;

[0084] The assessment module is used to perform risk assessments on precise electricity consumption data and obtain assessment results.

[0085] The screening module is used to screen for anomalies based on clustering and evaluation results, remove abnormal data, and obtain intermediate data. The intermediate data is smoothed by Gaussian filtering, and missing time period data is supplemented by interpolation based on the trend of adjacent time periods to obtain the target electricity consumption dataset.

[0086] The working principle and beneficial effects of the above technical solution are as follows: The clustering module divides complex electricity consumption data into distinctive clusters through pattern clustering, making the originally messy time-series data present clear patterns; the multi-layer risk assessment system of the evaluation module can not only identify anomalies at the data level, but also correlate equipment failures and user behavior risks; through the pattern classification of the clustering module and the risk identification of the evaluation module, the screening module can accurately remove abnormal data and obtain intermediate data; the intermediate data is smoothed by Gaussian filtering, and missing time period data is supplemented by interpolation based on the trend of adjacent time periods, which facilitates obtaining an accurate target electricity consumption dataset.

[0087] According to some embodiments of the present invention, a clustering module includes:

[0088] The segmentation module is used to divide the precise electricity consumption data into several time-series segments and extract the feature vector of each time-series segment.

[0089] The partitioning module is used to initially partition clusters based on feature vectors using the density clustering DBSCAN method, and then perform temporal alignment of the clusters based on the dynamic time warping algorithm.

[0090] The second determining module is used for:

[0091] After time alignment, boundary data points are determined; wherein, boundary data points are data points whose distance difference to the center of two clusters is less than a first preset threshold and whose local density is less than a second preset threshold of the average density within the cluster.

[0092] The feature similarity between the boundary data points and each cluster is calculated, and the affiliation information of the boundary data points is determined by the fuzzy membership function to obtain the clustering results.

[0093] The working principle and beneficial effects of the above technical solution are as follows: The segmentation module divides the precise electricity consumption data into several time-series segments and extracts the feature vector of each time-series segment, including statistical features, time-series features, and frequency features. Statistical features include average power and peak load; time-series features include load rise rate and peak / valley occurrence times; frequency features include electricity consumption fluctuation cycles. Based on the feature vectors, the density clustering DBSCAN method is used to initially divide the data into clusters. The dynamic time warping algorithm is then used to align the time-series segments within the same cluster, eliminating the interference of time offset on pattern similarity judgment and ensuring the temporal consistency of data within the cluster. Boundary data points are data that are easily misclassified. The feature similarity between boundary data points and each cluster is calculated, and the probability of belonging to different clusters is quantified using a fuzzy membership function. Data points with probabilities higher than a threshold are explicitly assigned to the corresponding clusters, forming the final clustering result. This avoids cluster ambiguity caused by boundary data and facilitates obtaining accurate clustering results.

[0094] According to some embodiments of the present invention, the evaluation module includes:

[0095] The module is used to construct a multi-layered risk assessment system. The bottom layer of the multi-layered risk assessment system consists of data risk factors, the middle layer consists of equipment risk factors, and the top layer consists of user risk factors. Data risk factors include the frequency of metering value jumps and the number of times parameters exceed limits. Equipment risk factors include the duration of chip temperature exceeding the standard and the response delay of communication modules. User risk factors include the load mutation rate, the duration of abnormal power consumption at night, and the three-phase imbalance.

[0096] The first calculation module is used to input accurate electricity consumption data into the multi-layer risk assessment system and use failure mode and effects analysis to calculate the failure score of risk factors.

[0097] The second calculation module is used to calculate the risk factor weights using the information entropy method.

[0098]

[0099] Among them, w k E represents the weight of the k-th risk factor. k E represents the information entropy of the k-th risk factor.m Let K be the information entropy of the m-th risk factor; K is the number of risk factors.

[0100] The third calculation module performs a weighted summation based on the failure scores and weights of the risk factors to obtain the risk value; based on the risk value and a preset risk level data table, it obtains the risk level as the evaluation result.

[0101] The working principle and beneficial effects of the above technical solution are as follows: A multi-layered risk assessment system is constructed based on building modules. After accurate electricity consumption data is input into the multi-layered risk assessment system, the first calculation module uses Failure Mode and Effects Analysis (FMEA) to quantify the risk level of each risk factor. By analyzing the occurrence frequency, impact severity, and detectability of each factor, a comprehensive assessment of the failure score of each risk factor is obtained. The second calculation module determines the weight of each risk factor using the information entropy method to reflect the differences in the impact of different factors on the overall risk. The third calculation module performs a weighted summation based on the failure scores and weights of the risk factors to obtain the risk value; based on the risk value and a preset risk level data table, the risk level is obtained as the assessment result. Constructing a multi-layered risk assessment system and using FMEA to calculate the failure score comprehensively considers the occurrence frequency, impact severity, and detectability of the risk, avoiding the one-sidedness of a single threshold judgment. The information entropy method for determining weights is based on the discrete characteristics of the data itself, reducing the bias of subjective weighting and facilitating accurate determination of the risk value, thereby accurately determining the assessment result.

[0102] According to some embodiments of the present invention, a behavior extraction module includes:

[0103] The third determining module is used for:

[0104] Based on the time-series feature decomposition method, the target electricity consumption dataset is decomposed into trend terms, periodic terms, and random terms; long-term features are extracted from the trend terms; and short-term features are extracted from the periodic terms.

[0105] Based on long-term and short-term characteristics, determine the user's initial electricity consumption characteristics;

[0106] User profiles are constructed based on users' initial electricity consumption characteristics. These user profiles are then compared with a profile database, and the user's differentiated electricity consumption characteristics are determined based on the comparison results.

[0107] The mining module is used to perform pattern mining on the target electricity consumption dataset to identify dynamic electricity consumption patterns; the dynamic electricity consumption patterns include burst patterns, periodic patterns and gradual patterns.

[0108] The working principle of the above technical solution is as follows: A trend term reflects long-term electricity consumption patterns; a periodic term reflects short-term recurring patterns. Long-term features are extracted from the trend term, including annual electricity consumption growth rate, seasonal fluctuation coefficient, and load rate distribution range. Short-term features are extracted from the periodic term, including: start and end points of daily peak hours, peak duration, peak-valley load ratio, and electricity consumption proportion for each time period. Based on the long-term and short-term features, preliminary user electricity consumption characteristics are determined. User profiles are constructed based on these preliminary characteristics. These profiles are compared with a profile database. The comparison results facilitate the determination of matching user profiles and their corresponding electricity consumption characteristics, enabling the acquisition of more comprehensive differentiated user electricity consumption characteristics and thus accurately determining user electricity consumption types, such as residential, commercial, and industrial electricity consumption. The data mining module performs pattern mining on the target electricity consumption dataset to identify dynamic electricity consumption patterns: sudden patterns are short-term surges / decreases in power consumption; periodic patterns are recurring electricity consumption behaviors within a fixed period; and gradual patterns are electricity consumption characteristics that change slowly over time.

[0109] The beneficial effects of the above technical solution are: it facilitates the accurate determination of users' differentiated electricity consumption characteristics and dynamic electricity consumption patterns.

[0110] According to some embodiments of the present invention, the generation module includes:

[0111] The control module is used to control target stratification based on differentiated electricity consumption characteristics;

[0112] The matching module is used to match differentiated control strategies based on dynamic power consumption patterns;

[0113] The fourth determination module is used to determine the adaptive metering control scheme based on the target stratification and differentiated control strategy.

[0114] The working principle and beneficial effects of the above technical solution are as follows: Based on differentiated electricity consumption characteristics, the control targets are stratified. Specifically: For industrial electricity, characterized by high load and strong cycles, priority is given to ensuring metering accuracy and reducing metering deviations caused by harmonic interference. For residential electricity, characterized by distinct peak and valley periods and short-term sudden changes, the metering sensitivity during peak and valley periods is optimized, such as avoiding metering delays during peak periods and reducing errors caused by sudden electricity consumption. For commercial electricity, characterized by random fluctuations and gradual increases, the focus is on dynamically tracking load change trends. Differentiated control strategies are matched according to dynamic electricity consumption patterns, including: for sudden patterns, a rapid response mechanism is activated; for periodic patterns, a cycle adaptation mechanism is activated; and for gradual patterns, a trend tracking mechanism is adopted. Based on the target stratification and differentiated control strategies, a metering adaptive control scheme is determined through a preset multi-dimensional control matrix, which facilitates the reduction of metering errors.

[0115] like Figure 3As shown, according to some embodiments of the present invention, the metering and control method of the artificial intelligence-based smart energy meter described above includes steps S1-S5:

[0116] S1. Obtain metering deviation data and multi-dimensional electricity consumption data;

[0117] S2. Correct the multi-dimensional electricity consumption data based on the pre-built dynamic compensation model and metering deviation data to obtain accurate electricity consumption data;

[0118] S3. Perform data analysis on the precise electricity consumption data to obtain the target electricity consumption dataset;

[0119] S4. Extract user-differentiated electricity consumption characteristics and dynamic electricity consumption patterns from the target electricity consumption dataset;

[0120] S5. Generate a metering adaptive control scheme based on differentiated electricity consumption characteristics and dynamic electricity consumption patterns, and execute the metering adaptive control scheme.

[0121] The beneficial effects of the above technical solution are as follows: By using a dynamic compensation model to correct multi-dimensional electricity consumption data in real time, it can adapt to metering deviations caused by dynamic factors such as power grid fluctuations, environmental changes, and component aging. This solves the problem that traditional static compensation mechanisms cannot cope with errors in complex scenarios, improving metering accuracy and ensuring the accuracy and reliability of electricity consumption data. Furthermore, by extracting differentiated user electricity consumption characteristics and dynamic electricity consumption patterns, the generated adaptive control scheme can match the electricity consumption habits of different users, avoiding the insufficient adaptability of traditional unified control strategies and facilitating improvements in the metering accuracy of smart meters.

[0122] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart energy meter based on artificial intelligence, characterized in that, include: The acquisition module is used to acquire metering deviation data and multi-dimensional electricity consumption data; The correction module is used to correct multi-dimensional electricity consumption data based on a pre-built dynamic compensation model and metering deviation data to obtain accurate electricity consumption data. The analysis module is used to perform data analysis on precise electricity consumption data to obtain the target electricity consumption dataset; The behavior extraction module is used to extract user-differentiated electricity consumption characteristics and dynamic electricity consumption patterns from the target electricity consumption dataset; The generation module is used to generate metering adaptive control schemes based on differentiated electricity consumption characteristics and dynamic electricity consumption patterns, and to execute the metering adaptive control schemes.

2. The smart energy meter based on artificial intelligence as described in claim 1, characterized in that, The acquisition module connects to the data monitoring module based on the power line carrier communication module and the LoRa communication module, and acquires metering deviation data and multi-dimensional electricity consumption data based on a sliding window.

3. The smart energy meter based on artificial intelligence as described in claim 1, characterized in that, The method for determining the dynamic compensation model in the correction module includes: Obtain historical measurement deviation data and corresponding target compensation values; Feature mining is performed on historical measurement deviation data to extract key influencing factors and error trends over a recent preset time period; key influencing factors include environmental parameters, power grid parameters, equipment parameters, and usage parameters; An initial compensation model is constructed. The initial compensation model includes a feedforward neural network and a timing compensator based on a GRU network. The feedforward neural network includes an input layer, two hidden layers, and an output layer. The number of nodes in the two hidden layers are 32 and 16, respectively. Key influencing factors are input into a feedforward neural network, which outputs a basic error compensation value; errors within a near-preset time period are input into a time series compensator, which outputs a dynamic correction value; the two are weighted and summed to obtain the predicted compensation value. The predicted compensation value is compared with the target compensation value, and the initial compensation model is iteratively trained based on the comparison results to obtain a trained dynamic compensation model.

4. The smart energy meter based on artificial intelligence as described in claim 3, characterized in that, The correction module includes: The first determining module is used to input the measurement deviation data into the pre-built dynamic compensation model to determine the dynamic compensation value; The compensation module is used to correct multi-dimensional electricity consumption data based on dynamic compensation values ​​to obtain accurate electricity consumption data.

5. The smart energy meter based on artificial intelligence as described in claim 1, characterized in that, The analysis module includes: The clustering module is used to perform pattern clustering on precise electricity consumption data to obtain clustering results; The assessment module is used to perform risk assessments on precise electricity consumption data and obtain assessment results. The screening module is used to screen for anomalies based on clustering and evaluation results, remove abnormal data, and obtain intermediate data. The intermediate data is smoothed by Gaussian filtering, and missing time period data is supplemented by interpolation based on the trend of adjacent time periods to obtain the target electricity consumption dataset.

6. The smart energy meter based on artificial intelligence as described in claim 5, characterized in that, The clustering module includes: The segmentation module is used to divide the precise electricity consumption data into several time-series segments and extract the feature vector of each time-series segment. The partitioning module is used to initially partition clusters based on feature vectors using the density clustering DBSCAN method, and then perform temporal alignment of the clusters based on the dynamic time warping algorithm. The second determining module is used for: After time alignment, boundary data points are determined; wherein, boundary data points are data points whose distance difference to the center of two clusters is less than a first preset threshold and whose local density is less than a second preset threshold of the average density within the cluster. The feature similarity between the boundary data points and each cluster is calculated, and the affiliation information of the boundary data points is determined by the fuzzy membership function to obtain the clustering results.

7. The smart energy meter based on artificial intelligence as described in claim 5, characterized in that, The evaluation module includes: The module is used to construct a multi-layered risk assessment system. The bottom layer of the multi-layered risk assessment system consists of data risk factors, the middle layer consists of equipment risk factors, and the top layer consists of user risk factors. Data risk factors include the frequency of metering value jumps and the number of times parameters exceed limits. Equipment risk factors include the duration of chip temperature exceeding the standard and the response delay of communication modules. User risk factors include the load mutation rate, the duration of abnormal power consumption at night, and the three-phase imbalance. The first calculation module is used to input accurate electricity consumption data into the multi-layer risk assessment system and use failure mode and effects analysis to calculate the failure score of risk factors. The second calculation module is used to calculate the risk factor weights using the information entropy method. The third calculation module performs a weighted summation based on the failure scores and weights of the risk factors to obtain the risk value; based on the risk value and a preset risk level data table, it obtains the risk level as the evaluation result.

8. The smart energy meter based on artificial intelligence as described in claim 1, characterized in that, The behavior extraction module includes: The third determining module is used for: Based on the time-series feature decomposition method, the target electricity consumption dataset is decomposed into trend terms, periodic terms, and random terms; long-term features are extracted from the trend terms; and short-term features are extracted from the periodic terms. Based on long-term and short-term characteristics, determine the user's initial electricity consumption characteristics; User profiles are constructed based on users' initial electricity consumption characteristics. These user profiles are then compared with a profile database, and the user's differentiated electricity consumption characteristics are determined based on the comparison results. The mining module is used to perform pattern mining on the target electricity consumption dataset to identify dynamic electricity consumption patterns; the dynamic electricity consumption patterns include burst patterns, periodic patterns and gradual patterns.

9. The smart energy meter based on artificial intelligence as described in claim 8, characterized in that, The generation module includes: The control module is used to control target stratification based on differentiated electricity consumption characteristics; The matching module is used to match differentiated control strategies based on dynamic power consumption patterns; The fourth determination module is used to determine the adaptive metering control scheme based on the target stratification and differentiated control strategy.

10. The metering and control method for an artificial intelligence-based smart energy meter as described in any one of claims 1-9, characterized in that, include: Acquire metering deviation data and multi-dimensional electricity consumption data; Based on the pre-built dynamic compensation model and metering deviation data, multi-dimensional electricity consumption data is corrected to obtain accurate electricity consumption data; Data analysis is performed on precise electricity consumption data to obtain the target electricity consumption dataset; Extract user-differentiated electricity consumption characteristics and dynamic electricity consumption patterns from the target electricity consumption dataset; A metering adaptive control scheme is generated based on differentiated electricity consumption characteristics and dynamic electricity consumption patterns, and then the metering adaptive control scheme is executed.

Images