Electric power measurement abnormity checking method, device and equipment based on artificial intelligence

Abstract

The invention relates to the technical field of intelligent electric meters, and discloses an artificial intelligence-based electric power measurement abnormity checking method, device and equipment, and the method comprises the steps: obtaining the electric power data of an electric energy meter, carrying out the normalization processing, and generating a time window sequence through a sliding window technology; and deep feature extraction is performed on the sequence in combination with a wavelet neural network, so that the capturing capability of the model on key information is enhanced. And inputting the extracted features into an electric power measurement anomaly detection model, wherein the model can accurately predict the electric power data of the next time step. By comparing the actual value with the predicted value, when the difference exceeds the preset threshold value, the system automatically judges that the metering of the electric energy meter is abnormal. According to the invention, the accuracy of electric power measurement and the real-time performance of anomaly detection are improved, and the reliability and maintenance efficiency of an electric power system are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of smart electric meters, and in particular to an artificial intelligence-based power metering abnormality verification method, device and equipment. Background Art

[0002] In traditional meter monitoring methods, the operating status of meters usually depends on regular manual calibration and on-site inspection. Although these traditional methods can ensure the accuracy of meters to a certain extent, they have significant disadvantages in practical applications, such as low efficiency, poor real-time performance, and inability to timely detect and correct errors in the operation of meters. Specifically, traditional meter monitoring relies on manual inspection and periodic calibration, and often ignores the sudden errors that may occur during non-periodic inspections, resulting in errors that cannot be detected and corrected in a timely manner. Therefore, in large-scale power metering systems, this monitoring method is difficult to meet the requirements of efficient, real-time, and accurate power metering.

[0003] With the development of smart meters and Internet of Things technologies, smart meters have gradually become the mainstream equipment for electricity metering. Smart meters can monitor power consumption data in real time and upload it to a cloud platform or monitoring center through the network. However, despite the advantages of real-time data collection and transmission, smart meters still have problems such as meter failure and abnormal operation. In particular, in the complex environment with a wide coverage of the power network, error detection and correction are still great challenges. Based on these backgrounds, more and more research is devoted to proposing meter monitoring methods based on data analysis and artificial intelligence.

[0004] In recent years, electric meter error monitoring methods based on artificial intelligence have emerged. Compared with traditional manual detection methods, smart meter operation error monitoring methods collect real-time operation data of electric meters and use data analysis, machine learning, neural network and other technologies to automatically and in real time analyze electric meter data, thereby discovering abnormal conditions in the operation of electric meters. Common monitoring methods in the prior art mainly include error detection methods based on models such as regression analysis, support vector machine (SVM), and decision tree. These methods can effectively analyze electric meter data, but they still face problems such as low prediction accuracy, high computational complexity, and difficulty in adapting to large-scale data. Summary of the invention

[0005] In view of this, the present application provides an artificial intelligence-based power metering anomaly verification method to solve the technical problems of low accuracy and high computational complexity in the prior art of power metering anomaly verification.

[0006] According to a first aspect of the present application, a method for checking abnormality of power metering based on artificial intelligence is provided, comprising: Acquire power data of the power meter, wherein the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; Normalizing the acquired power data; Based on a sliding window, segmenting the normalized power data to generate at least one time window sequence; Extracting features from the time window sequence using a wavelet neural network to generate an input sequence; Inputting the input sequence into the power metering anomaly detection model to obtain a predicted value of power data at the next time step; Calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold, it is determined that the electric energy meter has a metering abnormality.

[0007] According to a second aspect of the present application, there is provided an artificial intelligence-based power metering abnormality verification device, comprising: An acquisition module, used for acquiring power data of the electric energy meter, wherein the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; A preprocessing module, used for normalizing the acquired power data; An input processing module, configured to segment the normalized power data based on a sliding window to generate at least one time window sequence; A feature extraction module, used for extracting features from the time window sequence through a wavelet neural network to generate an input sequence; A prediction module, used for inputting the input sequence into the power metering anomaly detection model to obtain a predicted value of power data at the next time step; The detection module is used to calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold value. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold value, it is judged that the electric energy meter has a metering abnormality.

[0008] According to a third aspect of the present application, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned artificial intelligence-based power metering anomaly verification method when executing the computer program.

[0009] According to a fourth aspect of the present application, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned artificial intelligence-based electricity metering anomaly verification method are implemented.

[0010] By means of the above technical scheme, the present application provides an artificial intelligence-based power metering anomaly verification method, device, and equipment, which obtains the power data of the power meter, performs normalization processing, and generates a time window sequence using sliding window technology. The deep feature extraction of the sequence is combined with the wavelet neural network, which enhances the model's ability to capture key information. The extracted features are input into the power metering anomaly detection model, which can accurately predict the power data of the next time step. By comparing the actual value with the predicted value, when the difference exceeds the preset threshold, the system will automatically determine that the power meter measurement is abnormal. The present invention improves the accuracy of power metering and the real-time performance of anomaly detection, and optimizes the reliability and maintenance efficiency of the power system.

[0011] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application and to implement it in accordance with the contents of the specification, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 A schematic diagram of an application scenario of an artificial intelligence-based power metering abnormality verification method provided in an embodiment of the present application is shown; Figure 2 A schematic diagram of a flow chart of an artificial intelligence-based power metering abnormality verification method provided in an embodiment of the present application is shown; Figure 3 A schematic diagram of the process of training the power metering anomaly detection model provided in an embodiment of the present application is shown; Figure 4 A schematic diagram of a process for optimizing the training parameters of the power metering anomaly detection model using the fruit fly optimization algorithm provided in an embodiment of the present application is shown; Figure 5 A schematic diagram of the structure of an artificial intelligence-based power metering anomaly verification device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0013] The specific implementation of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0014] An artificial intelligence-based power metering abnormality verification method provided by an embodiment of the present invention can be applied to Figure 1 The power metering anomaly verification system shown can be applied to smart grids and includes power meters, data acquisition equipment and data analysis servers to achieve real-time monitoring and analysis of power meter operation errors. Among them, the power meter is responsible for real-time measurement and recording of power consumption data, including current, voltage, power factor and other parameters; on a group of power meters or nearby edge computing devices, lightweight data processing and preliminary analysis modules can be deployed to process and filter key data in real time and reduce the amount of data transmission; the data acquisition equipment is responsible for collecting data from multiple power meters and transmitting the data to the central server; the data analysis server stores a large amount of historical data and real-time data transmitted from the power meters and data acquisition equipment; the power metering anomaly detection model provided by the embodiment of the present invention is deployed on the data analysis server, which performs real-time analysis on the operation data of the power meter, detects potential anomalies and errors, and executes a method based on artificial intelligence provided by the present invention. The invention discloses a method for checking abnormality of electric power metering, which comprises the steps of obtaining electric power data of an electric power meter; normalizing the obtained electric power data; segmenting the normalized electric power data based on a sliding window to generate at least one time window sequence; extracting features of the time window sequence through a wavelet neural network to generate an input sequence; inputting the input sequence into an electric power metering abnormality detection model to obtain a predicted value of electric power data at the next time step; calculating whether the difference between the actual value of the electric power data of the electric power meter and the predicted value of the electric power data at the next time step is greater than a preset threshold value, and judging that the electric power meter has a metering abnormality fault when the absolute value of the difference between the actual value of the electric power data of the electric power meter and the predicted value of the electric power data at the next time step is greater than the preset threshold value.

[0015] The present invention is described in detail below through specific embodiments.

[0016] Embodiment 1: like Figure 2 As shown, an artificial intelligence-based power metering abnormality verification method provided in an embodiment of the present invention includes: Step 201, obtaining power data of the electric energy meter; Wherein, the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; Step 202: normalize the acquired power data; Wherein, step 202 specifically includes: Step a: Calculate the minimum and maximum values ​​of each power parameter in the power data; Step b: Subtract the minimum value from the value of each power parameter; Step c: Divide the result of step b by the difference between the maximum and minimum values ​​so that the normalized value is in the interval [0, 1].

[0017] This step normalizes the various indicators of the power data so that data of different dimensions and ranges can be compared and processed on the same scale, improving the efficiency and accuracy of model training.

[0018] Step 203: segment the normalized power data based on the sliding window to generate at least one time window sequence; The sliding window technology is used to segment the normalized power data and generate a time window sequence. The present invention can capture the changing characteristics of power data in different time periods. This dynamic analysis method helps to identify short-term fluctuations and long-term trends in the power metering process.

[0019] Step 204: extract features from the time window sequence through a wavelet neural network to generate an input sequence; Among them, by extracting features from the time window sequence through a wavelet neural network, representative and distinguishing time-frequency features are extracted from the power data in the embodiment of the present invention. These features provide rich information for power metering anomaly detection and enhance the recognition ability of the model.

[0020] Step 205: input the input sequence into the power metering anomaly detection model to obtain the power data prediction value of the next time step; Step 206, calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold value. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold value, it is determined that the electric energy meter has a metering abnormality.

[0021] Embodiment 1 of the present invention provides an artificial intelligence-based power metering anomaly verification method, which obtains power data of an electric energy meter; normalizes the acquired power data; segments the normalized power data based on a sliding window to generate at least one time window sequence; extracts features of the time window sequence through a wavelet neural network to generate an input sequence; inputs the input sequence into an electric energy metering anomaly detection model to obtain a predicted value of the power data for the next time step; calculates whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data for the next time step is greater than a preset threshold value, and when the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data for the next time step is greater than the preset threshold value, it is determined that the electric energy meter has a metering abnormality fault.

[0022] Embodiment 2: The step before step 204 includes a training step of the power metering anomaly detection model, such as Figure 3 As shown, specifically including: Step 301: Acquire historical power data of the electric energy meter as training data; Step 302: normalize the acquired training data, and segment the normalized training data based on a sliding window to generate at least one time window training sequence; Step 303: extract features from the time window training sequence through a wavelet neural network to generate an input training sequence; Step 304: Input the input training sequence into the LSTM model, and optimize the network weights through the loss function and the back propagation algorithm to generate a power metering anomaly detection model.

[0023] The model training of the second embodiment of the present invention collects the historical power data of the electric energy meter through step 301 to ensure the comprehensiveness of the training data. In step 302, these data are normalized and divided into time window training sequences through sliding window technology to capture the dynamic characteristics of time series data. Step 303 uses wavelet neural network to extract features from these sequences, thereby enhancing the model's ability to recognize time-frequency features. Finally, in step 304, the extracted features are input into the LSTM model, and the network weights are optimized through loss function and back propagation algorithm to construct an efficient power metering anomaly detection model. This process not only improves the accuracy of anomaly detection, but also enhances the adaptability and prediction ability of the model to complex time series data, realizes the accurate identification of power metering anomalies, and provides a strong guarantee for the stable operation of the power system.

[0024] Embodiment three: In view of the rapid and large-span changes in the operating indicators of the electric energy meter, the LSTM fixed prediction window in the prior art cannot adapt to the error analysis task of the electric energy meter. Therefore, a dynamic sliding window is used in Example 2 of the present invention. This sliding window can effectively handle the error analysis problem of the electric energy meter with time characteristics and efficiently capture the indicator changes during the operation of the electric energy meter. Step 203 of Example 1 of the present invention includes: Step 203-1, starting from the power data, select an initial window; The initial window contains data points of time steps; Step 203-2: According to the volatility of power data, Dynamically resize the window.

[0025] in, is the initial window size, The window size at the nth time step, The weighted value of the standard deviation or change rate of each indicator data of the power data in the nth time step.

[0026] Embodiment 4: In order to adapt to the input form of the power metering anomaly detection model, a mapping transformation is required. In the fourth embodiment of the present invention, it is proposed to map the features in the time window sequence to a new feature space to form a mapping sequence. Step 204 of the first embodiment of the present invention includes: Step 204-1: By according to Perform mapping transformation to obtain mapping sequence , where k is the kth feature of the time window sequence, , m is the number of features of the time window sequence, i is the i-th feature of the mapping sequence, , q is the number of features of the mapping sequence, t is the time step, for unbalanced function, is the weight function from the kth unit in the input layer to the kth unit in the hidden layer, and are the scale and displacement of the wavelet function, is the connection weight from the i-th unit in the hidden layer to the j-th unit in the output layer; Step 204-1: Map the sequence as input sequence.

[0027] Embodiment 4 of the present invention maps the features in the time window sequence to a new feature space through mapping transformation to form a mapping sequence. This transformation can convert the original features into a form that is more conducive to model learning and improve the expressiveness of the features. The time window sequence is processed using wavelet transform, where and Represent the scale and displacement parameters of the wavelet function respectively. The introduction of wavelet transform enables the model to capture local features and changes in time series data, which is crucial for identifying abnormal patterns in power metering. Unbalanced Sigmoid function: By applying the unbalanced Sigmoid function, the nonlinear expression ability of the model is increased, which helps to deal with complex nonlinear relationships, thereby improving the prediction performance of the model.

[0028] Embodiment five: Based on the first embodiment, a weighted attention mechanism is introduced into the output of the power metering anomaly detection model, specifically including: The weighted context vector Input to the output layer as the final prediction output, where , is the attention weight at time step t, , , is the probability distribution weight, the activation function tanh is the hyperbolic tangent function, is the current output of the LSTM neural network, w is the weight of the neuron, and g is the offset of the neuron.

[0029] In the fifth embodiment of the present invention, a weight function is introduced during the mapping process to adjust the feature mapping from the input layer to the hidden layer, which helps the model learn more important feature representations. The output of the hidden layer is mapped to the output layer by connecting weights, and these weights are continuously optimized during the training process to minimize the prediction error. Through the above feature mapping and transformation, the model can better generalize to unseen data and improve the prediction accuracy of new data. The mapping sequence is used as an input sequence, so that the LSTM model can more effectively identify and predict abnormal situations in power metering, thereby improving the stability and reliability of the power system.

[0030] Embodiment six: Based on the first embodiment, the training parameters of the power metering anomaly detection model are optimized by the fruit fly optimization algorithm. The global optimization of the LSTM model parameters is performed by the fruit fly optimization algorithm. Combined with the dynamic update mechanism, social effect, food source simulation and local search strategy, the present invention significantly improves the accuracy of power metering anomaly detection and the efficiency of model training, and realizes efficient guarantee of stable operation of the power system. Figure 4 As shown, including: Step a, randomly initialize the position and speed of the fruit fly group, each fruit fly represents a potential solution of the power metering anomaly detection model; Step b, calculating the fitness function of each fruit fly solution, wherein the fitness function is based on the prediction error of the LSTM model; Step c, updating the position and speed of the fruit fly according to the fitness of the fruit fly; Step d: Introducing social effects to allow fruit flies to share information; Step e, simulating the fruit fly's search for a food source, updating the location of the food source to guide the fruit fly to move to a better solution; Step f: Based on the global search, introduce a local search strategy to fine-tune the current optimal solution; Step g, repeat steps b to f until a stopping condition is met, such as reaching a maximum number of iterations or a fitness threshold; Step h: Use the fruit fly algorithm to optimize the parameters of the LSTM network, including weights and biases; Step i: Retrain the LSTM network based on the parameters optimized by the fruit fly algorithm.

[0031] Further, as Figures 2 to 4 In the specific implementation of the method, an artificial intelligence-based power metering abnormality verification device is provided in the embodiment of the present invention, such as Figure 5 As shown, the device comprises: An acquisition module 510 is used to acquire power data of an electric energy meter, wherein the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; A preprocessing module 520, configured to perform normalization processing on the acquired power data; An input processing module 530, configured to segment the normalized power data based on a sliding window to generate at least one time window sequence; A feature extraction module 540 is used to extract features from the time window sequence through a wavelet neural network to generate an input sequence; A prediction module 550, configured to input the input sequence into a power metering anomaly detection model to obtain a predicted value of power data at the next time step; The detection module 560 is used to calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold value. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold value, it is determined that the electric energy meter has a metering abnormality.

[0032] In an embodiment of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of implementing the method for checking abnormality of power metering based on artificial intelligence include: Acquire power data of the power meter, wherein the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; Normalize the acquired power data; Based on the sliding window, the normalized power data is segmented to generate at least one time window sequence; The feature extraction of the time window sequence is performed through the wavelet neural network to generate the input sequence; Input the input sequence into the power metering anomaly detection model to obtain the predicted value of power data at the next time step; Calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold, it is determined that the electric energy meter has a metering abnormality.

[0033] In an embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: Acquire power data of the power meter, wherein the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; Normalizing the acquired power data; Based on the sliding window, the normalized power data is segmented to generate at least one time window sequence; Extracting features from the time window sequence using a wavelet neural network to generate an input sequence; Input the input sequence into the power metering anomaly detection model to obtain the predicted value of power data at the next time step; Calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold, it is determined that the electric energy meter has a metering abnormality.

[0034] It should be noted that the above embodiments only use smart meters as examples to illustrate the principles and implementation steps of the embodiments of the present invention, and do not specifically limit the actual application scenarios. For example, power meters, power factor meters, harmonic analyzers, relay protection instruments, switch quantity acquisition instruments, power quality analyzers, etc., the technical solution of the present invention can also be applied to metering fault prediction of other power grid-related instruments. For the functions or steps that can be implemented by computer-readable storage media or computer devices, please refer to the aforementioned method embodiments. To avoid repetition, they will not be described one by one here.

[0035] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0036] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0037] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for checking abnormality of power metering based on artificial intelligence, characterized in that: include: Acquire power data of the power meter, wherein the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; Normalizing the acquired power data; Based on a sliding window, segmenting the normalized power data to generate at least one time window sequence; Extracting features from the time window sequence through a wavelet neural network to generate an input sequence; Inputting the input sequence into the power metering anomaly detection model to obtain the power data prediction value of the next time step; Calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold, it is determined that the electric energy meter has a metering abnormality.

2. The artificial intelligence-based power metering abnormality verification method according to claim 1 is characterized in that: The step of inputting the input sequence into the power metering anomaly detection model to obtain the power data prediction value of the next time step includes a training step of the power metering anomaly detection model, which specifically includes: Obtain historical power data of the electric energy meter as training data; Normalizing the acquired training data, and segmenting the normalized training data based on a sliding window to generate at least one time window training sequence; Extracting features from the time window training sequence through a wavelet neural network to generate an input training sequence; The input training sequence is input into the LSTM model, and the network weight is optimized through the loss function and the back propagation algorithm to generate an electric power metering anomaly detection model.

3. The artificial intelligence-based power metering abnormality verification method according to claim 1 is characterized in that: The step of normalizing the acquired power data comprises: Step a: Calculate the minimum and maximum values ​​of each power parameter in the power data; Step b: Subtract the minimum value from the value of each power parameter; Step c: Divide the result of step b by the difference between the maximum and minimum values ​​so that the normalized value is in the interval [0, 1].

4. The method for checking abnormality of power metering based on artificial intelligence according to claim 1 is characterized in that: The step of segmenting the normalized power data based on the sliding window to generate at least one time window sequence includes: Starting from the power data, an initial window is selected, wherein the initial window contains data points of time steps; According to the volatility of the power data, Dynamically adjust the window size, where is the initial window size, The window size at the nth time step, The weighted value of the standard deviation or change rate of each indicator data of the power data in the nth time step.

5. The method for checking abnormality of power metering based on artificial intelligence according to claim 1 is characterized in that: The step of extracting features from the time window sequence through a wavelet neural network to generate an input sequence comprises: By dividing the time window sequence According to the formula , perform mapping transformation to obtain mapping sequence , where k is the kth feature of the time window sequence, , m is the number of features of the time window sequence, i is the i-th feature of the mapping sequence, , q is the number of features of the mapping sequence, t is the time step, for unbalanced function, is the weight function from the kth unit in the input layer to the kth unit in the hidden layer, and are the scale and displacement of the wavelet function, is the connection weight from the i-th unit in the hidden layer to the j-th unit in the output layer; The mapping sequence as input sequence.

6. The artificial intelligence-based power metering abnormality verification method according to claim 2 is characterized in that: A weighted attention mechanism is introduced into the output of the power metering anomaly detection model, specifically including: The weighted context vector Input to the output layer as the final prediction output, where , is the attention weight at time step t, , , is the probability distribution weight, the activation function tanh is the hyperbolic tangent function, is the current output of the LSTM neural network, w is the weight of the neuron, and g is the offset of the neuron.

7. The artificial intelligence-based power metering abnormality verification method according to claim 2 is characterized in that: The step of inputting the input training sequence into the LSTM model and optimizing the network weights through the loss function and the back propagation algorithm to generate the power metering anomaly detection model also includes: Step a, randomly initialize the position and speed of the fruit fly group, each fruit fly represents a potential solution of the power metering anomaly detection model; Step b, calculating the fitness function of each fruit fly solution, wherein the fitness function is based on the prediction error of the LSTM model; Step c, updating the position and speed of the fruit fly according to the fitness of the fruit fly; Step d: Introducing social effects, allowing fruit flies to share information, such as optimal locations; Step e, simulating the fruit fly's search for a food source, and updating the location of the food source to guide the fruit fly to move to a better solution; Step f: Based on the global search, introduce a local search strategy to fine-tune the current optimal solution; Step g, repeat steps b to f until a stopping condition is met, such as reaching a maximum number of iterations or a fitness threshold; Step h: Use the fruit fly algorithm to optimize the parameters of the LSTM network, including weights and biases; Step i: Retrain the LSTM network based on the parameters optimized by the fruit fly algorithm.

8. An artificial intelligence-based power metering abnormality verification device, characterized in that: include: An acquisition module, used for acquiring power data of the electric energy meter, wherein the power data is time series data, and the power data includes current, voltage, power factor and frequency of a preset time step; A preprocessing module, used for normalizing the acquired power data; An input processing module, configured to segment the normalized power data based on a sliding window to generate at least one time window sequence; A feature extraction module, used to extract features from the time window sequence through a wavelet neural network to generate an input sequence; A prediction module, used for inputting the input sequence into the power metering anomaly detection model to obtain a predicted value of power data at the next time step; The detection module is used to calculate whether the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than a preset threshold value. When the absolute value of the difference between the actual value of the power data of the electric energy meter and the predicted value of the power data at the next time step is greater than the preset threshold value, it is judged that the electric energy meter has a metering abnormality.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the artificial intelligence-based power metering anomaly verification method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the power metering anomaly verification method based on artificial intelligence as described in any one of claims 1 to 7 are implemented.

Images