Electric energy metering information transmission system and method

Abstract

The application relates to the technical field of electric energy metering information transmission, and discloses an electric energy metering information transmission system and method. Through data acquisition and preprocessing, voltage, current and power signals are collected, a normalized time sequence matrix is generated, and the feature expression capability is enhanced through nonlinear transformation to solve the signal complexity problem. Feature extraction and compression are combined with sparse processing to extract key feature information, reduce data dimension and redundancy, reduce storage and transmission pressure, and improve transmission efficiency. A dynamic transmission strategy is based on adaptive adjustment of frequency and priority based on network state, and a verification mechanism is combined to enhance transmission reliability. Decoding and reconstruction use an inverse reconstruction algorithm to recover signal features, ensuring accuracy and data availability under high compression ratio. Data analysis and anomaly detection combine high-quality features and dynamic algorithms to identify abnormal values and abnormal fluctuations in real time, improving the adaptability of the system in complex environments.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric energy metering information transmission, in particular to an electric energy metering information transmission method. BACKGROUND

[0002] With the rapid development of smart grid, electric energy metering information transmission system gradually becomes an important part of the power system. Accurate collection, compression, transmission and processing of electric energy metering information are of great significance to the efficiency and stability of power grid operation and the improvement of user service experience. However, there are still many problems in the actual application of existing electric energy metering information transmission technology, especially in the aspects of data compression, transmission efficiency, data integrity and abnormal processing, which restricts the overall performance of the system.

[0003] In the prior art, whether it is wireless communication or wired communication, the rapid increase of data volume easily leads to low transmission efficiency. Especially in the high-frequency sampling scene, the transmission of a large amount of raw data poses a serious challenge to bandwidth and delay. The processing capacity of conventional compression algorithms for high-dimensional and multi-source data is limited, and the redundant information in the data cannot be fully utilized. The decompression process may cause additional delay. The general encryption method has high computational complexity, which easily causes performance bottleneck to resource-limited electric energy metering devices such as embedded systems. In addition, the traditional method has limited ability to resist data tampering and forgery. The current cloud analysis mode is difficult to feedback the analysis results in real time, and relies on complex deep learning or statistical modeling methods in abnormal data detection, resulting in high processing delay.

[0004] In view of the above problems, the present application provides a new electric energy metering information transmission system and method based on data auto-encoding technology, aiming to solve the pain points in the prior art and significantly improve the efficiency, security and real-time performance of data transmission and analysis. SUMMARY

[0005] The present application provides an electric energy metering information transmission system and method, which solves the problems mentioned in the background.

[0006] The present application provides the following technical solution: an electric energy metering information transmission method, comprising:

[0007] S1, data collection and preprocessing;

[0008] The voltage measurement module is used to measure the voltage signal, denoted as V(t);

[0009] The current sensor is used to measure the current signal, denoted as I(t);

[0010] The power calculation module is used to calculate the real-time power signal, denoted as P(t);

[0011] Where t is the timestamp.

[0012] Time series organization of voltage signals, current signals and power signals to form a matrix:

[0013] ;

[0014] Wherein, X(t) is a matrix containing voltage signals, current signals and power signals;

[0015] Normalization of signals:

[0016] ;

[0017] Wherein, X ij is the jth value of the ith signal in the matrix X(t); is the mean of the ith signal; is the standard deviation of the ith signal;

[0018] Using nonlinear transformation to enhance the expression ability of signal features:

[0019] ;

[0020] Wherein, α ij , β ij and g ij are dynamic change parameters of signals; tanh() and sin() are hyperbolic tangent and sine functions respectively, which are used for nonlinear transformation;

[0021] S2, feature extraction and compression;

[0022] S3, data compression and sparsification;

[0023] S4, data transmission;

[0024] S5, data decoding and reconstruction;

[0025] S6, data analysis and anomaly detection.

[0026] Optionally, the feature extraction and compression specifically includes:

[0027] Extracting time domain and frequency domain features of each signal:

[0028] ;

[0029] Wherein, f i is the time domain feature of the ith signal; T is the time length of the signal;

[0030] Principal component analysis of the extracted features:

[0031] ;

[0032] wherein, U, and V are the matrix after SVD decomposition; F' is the compressed feature vector matrix;

[0033] Activate the compressed feature vector:

[0034] ;

[0035] wherein, F is the feature vector matrix; tanh() is the hyperbolic tangent function.

[0036] Optionally, the data compression and sparsification specifically includes:

[0037] Sparsify the compressed feature:

[0038] ;

[0039] wherein, F' i is the i-th value in the feature matrix; l is a sparsification control parameter, used to adjust the intensity of sparsification; β is the index of sparsification, β>1 is used to enhance sparseness;

[0040] as an important feature if it satisfies the condition ; otherwise, as an unimportant feature, and set the unimportant feature to 0:

[0041] ;

[0042] wherein, F final i is the screened feature value; θ is the threshold value, used to determine whether the feature is retained;

[0043] Output the final sparse feature matrix F final , which will be used for transmission.

[0044] Optionally, the data transmission specifically includes:

[0045] S41, generate a feature encoding matrix F encoded ;

[0046] From the original input signal matrix X(t), obtain the sparse matrix F sparse through the sparse feature extraction algorithm:

[0047] ;

[0048] wherein, X(t) is the signal matrix; A is the feature basis matrix; l is the sparsity regularization parameter;

[0049] Fsparse The encryption encoding is performed to generate an encoded feature matrix F encoded :

[0050] ;

[0051] wherein, is an encryption mapping function; () is a feature checking function, specifically:

[0052] ;

[0053] wherein, W1 is the weight matrix of the checking function; b1 is the bias vector;

[0054] S42, network state monitoring;

[0055] The network state is monitored in real time to obtain the current delay , bandwidth B(t) and lost packet P(t);

[0056] S43, transmission rate calculation;

[0057] According to the network state, the data transmission rate R(t) is adaptively adjusted:

[0058] ;

[0059] wherein, R min is the minimum transmission rate; is the delay difference of the current network; is the maximum delay allowed by the system; is the delay threshold; α, β and are parameters for adjusting the transmission rate;

[0060] S44, generating data packets;

[0061] The length of the maximum data packet transmitted by the network is obtained, denoted as L max ;

[0062] F encoded is divided into multiple data packets:

[0063] ;

[0064] wherein, P i is the i-th data packet; i represents the index of the data packet; L max represents the maximum length of each data packet;

[0065] Each column of the data packet corresponds to a feature component, and each row of the data packet corresponds to a signal at a specific time step;

[0066] If the last data packet is not enough L max , padding check bit 0 is filled in;

[0067] S45, data packaging and coding;

[0068] F encoded is divided into multiple small data packets P i according to the transmission rate R(t), and redundant check bits are added to each data packet;

[0069] Wherein is the redundant check bit function of the data packet, specifically:

[0070] ;

[0071] Wherein, P i,k is the kth data unit in the data packet P i ; is the weight coefficient, which is related to the length of the data packet; mod(•,2) is the modulo 2 operation, which is used to generate binary check bits;

[0072] S46, data sending and checking;

[0073] Send P i at the rate of R(t), and check the data integrity in real time at the receiving end:

[0074] ;

[0075] If the check fails, data retransmission is performed.

[0076] Optionally, the data decoding and reconstruction specifically includes:

[0077] S51, data receiving and checking;

[0078] For each received data packet P i , first perform integrity check:

[0079] ;

[0080] Wherein, P i is the ith received data packet; is the check bit generated by the check function;

[0081] If is 0, it means that the data packet has not been tampered with, and the data is complete; otherwise, the receiving end requests retransmission of the data packet;

[0082] S52, data decryption;

[0083] Use the data decryption formula to decrypt the received data packet Pi Decrypt(F

[0084] ;

[0085] wherein, F decoded i is the decrypted data matrix; Decrypt() is a decryption function, specifically:

[0086] ;

[0087] wherein, b1 is a bias term; W -1 1 is the inverse matrix of the check function.

[0088] S53, data reconstruction and splicing;

[0089] All decrypted data segments F decoded i are spliced to reconstruct a complete feature matrix F decoded :

[0090] ;

[0091] wherein, ∪ represents a splicing operation, and the decryption results of all data packets are spliced in order to form a complete feature matrix;

[0092] S54, data analysis and reconstruction;

[0093] The original data features F sparse are restored through reverse transformation:

[0094] ;

[0095] wherein, R() represents a reverse reconstruction function, and the decrypted feature matrix is restored to the original sparse feature matrix.

[0096] Optionally, the reverse reconstruction function specifically includes:

[0097] The design of the reverse reconstruction function R() is specifically:

[0098] S541, input variables;

[0099] The input variables include F encoded , , and P i ;

[0100] The F encoded is a data matrix after encoding; is a check function used in the encoding process; P i is a data packet;

[0101] S542, output variable;

[0102] Output variable F reconstructed ;

[0103] The F reconstructed is the feature matrix after inverse reconstruction;

[0104] S543, data inverse decompression;

[0105] Using inverse transformation matrix T -1 to perform data inverse decompression, the data after inverse decompression is recorded as F compressed :

[0106] ;

[0107] S544, data inverse decryption;

[0108] The data after inverse decompression is decrypted by inverse decryption algorithm, and the data matrix after inverse decryption is recorded as F decrypted :

[0109] ;

[0110] S545, data inverse verification;

[0111] ;

[0112] Wherein, () is the inverse verification function, F reconstructed is the reconstructed data matrix.

[0113] Optionally, the data analysis and anomaly detection specifically includes:

[0114] Using adaptive transformation method to construct feature space of data:

[0115] ;

[0116] Wherein, F transformed is the transformed feature matrix; T is the adaptive transformation matrix;

[0117] Using weighted locality distance measure to evaluate the abnormal degree of data points:

[0118] ;

[0119] Wherein, D(x i ) is the abnormality measure of the i-th data point; x i is the transformed feature matrix of the i-th data point; x j is the transformed feature vector of the j-th data point; is the weighting coefficient;

[0120] Setting a dynamic threshold T d To determine abnormal data points:

[0121] ;

[0122] Where, is the mean of the weighted local distance of all data points; is the standard deviation of the weighted local distance of all data points; α is a hyperparameter that controls the sensitivity of the threshold;

[0123] According to the threshold T d Label abnormal data points;

[0124] Set a label vector A, whose elements are 1 or 0, indicating whether the data point is abnormal:

[0125] ;

[0126] Where, A i is the i-th element in the label vector; 1 indicates abnormal; 0 indicates normal;

[0127] Repair abnormal data using a repair method based on local patterns, abnormal data points x i Repair through the following repair algorithm:

[0128] ;

[0129] Where, x repaired i is the repaired data point; N i is the domain of data point x i , containing similar normal data points to x i ; is the weight calculated based on the similarity between data points.

[0130] A system for implementing the power metering information transmission method, comprising:

[0131] Voltage measurement module: for measuring voltage signal;

[0132] Current sensor: for measuring current signal;

[0133] Power calculation module: for calculating real-time power signal;

[0134] Network state monitoring module: real-time monitoring of network state, obtaining current delay, bandwidth and loss packet.

[0135] The present application has the following advantages:

[0136] By data acquisition and preprocessing, the problem of complex power signal and insufficient feature expression ability is solved, effectively improving the expression and analysis ability of the signal. Voltage measurement module, current sensor and power calculation module are used to collect voltage signal, current signal and real-time power signal respectively, and time series organization is formed into a matrix. Then, the matrix is normalized to solve the problem of difficult unified processing of signals in different dimensions and amplitude ranges. At the same time, through nonlinear transformation (using hyperbolic tangent and sine function combined with dynamic change parameter to enhance the nonlinear feature expression ability of the signal), a high-quality foundation is laid for subsequent feature extraction and analysis. Through feature extraction and compression, data sparsification steps, the problem of high data dimension and low transmission efficiency is solved, effectively reducing the pressure of data storage and transmission. In the feature extraction and compression process, important feature information in the power signal is extracted, and redundant information is discarded. Then, through the sparsification process, the dimension and redundancy of the data are further reduced, the volume of the data to be transmitted is reduced, the data transmission efficiency is optimized, and the system's adaptability to large-scale users is improved. Through the data transmission step, the problem of data loss and unstable transmission in the process of power information transmission is solved, and the reliability of data transmission is improved. Using the data packet processed by feature extraction and sparsification, combined with dynamic transmission strategy, such as adjusting the frequency and priority of data transmission based on network status, the risk of network congestion is effectively reduced. At the same time, the checking mechanism in the transmission process guarantees the integrity and accuracy of the data. Through the data decoding and reconstruction step, the problem of insufficient feature reconstruction accuracy under high compression ratio is solved, ensuring the usability and accuracy of the decoded data. In data decoding and reconstruction, the feature information extracted in the compression process is used to recover the original signal features through inverse reconstruction algorithm. This process combines the checking mechanism to effectively improve the reconstruction accuracy of the decoded data, providing reliable basic data for subsequent analysis. Through the data analysis and anomaly detection step, the problem of difficult detection of abnormal values in power metering data and insufficient real-time analysis efficiency is solved, improving the intelligent level of the system. Using the extracted high-quality signal features, combined with dynamic analysis algorithm, the data is processed and detected in real time, which can quickly identify abnormal conditions in the power metering process, such as signal noise and power abnormal fluctuation, effectively reducing the false positive and false negative rates, and enhancing the system's adaptability to complex power environment.

[0137] By extracting the time domain and frequency domain features of each signal, the limitations of single-dimensional features in fully reflecting signal characteristics are addressed, effectively improving the comprehensiveness and accuracy of signal description. During the feature extraction process, the time domain features of the signal are calculated, fully reflecting the changing trend of the signal in the time dimension. At the same time, through frequency domain feature analysis, the frequency distribution characteristics of the signal are revealed. The combination of time domain and frequency domain features not only retains the time dynamic information of the signal, but also enhances the ability to characterize the periodicity, energy distribution and other characteristics of the signal, laying a foundation for subsequent data compression and analysis. Through principal component analysis of the extracted features, the problem of high-dimensional feature data redundancy and high computational complexity is solved, effectively reducing the dimension of the data and optimizing the computational efficiency. Principal component analysis extracts the main feature components through singular value decomposition, and the compressed feature matrix only retains the most representative part of the signal. This step not only reduces the redundancy of the data, but also improves the storage and transmission efficiency, while avoiding the interference of too many irrelevant features on the analysis results. By activating the compressed feature vector, the problem of insufficient signal feature expression ability is solved, effectively enhancing the non-linear expression ability of the features. Using the hyperbolic tangent function to activate the feature vector matrix, the non-linear expression ability of the signal is further improved. This process can highlight the dynamic characteristics and weak features hidden in complex signals, making the signal features more distinguishable and providing a more reliable data basis for subsequent analysis and anomaly detection.

[0138] By sparsifying the compressed features, the problem of high data redundancy is solved, effectively improving the data transmission efficiency. Sparse processing sparsifies the feature matrix through a formula, where the sparse control parameter and the sparse index β are used to adjust the sparsity intensity and sparse effect. A larger l value compresses smaller features more, and β > 1 enhances the sparsity. This process significantly reduces irrelevant or redundant feature information in the data, reducing the complexity of subsequent processing and transmission. By filtering feature values based on a threshold, the problem of insufficient feature filtering accuracy is solved, optimizing the quality of the final feature matrix. Using a conditional filtering formula to filter the sparsified features, where q is the threshold, ensures that only feature values that significantly contribute to signal expression are retained. Setting unimportant features to zero effectively reduces the data volume while retaining important feature information, improving the transmission efficiency and analysis accuracy of the feature matrix. By outputting the final sparse feature matrix, the problem of quality degradation in compressed data in traditional methods is solved, significantly improving the reliability and effectiveness of feature transmission. The final sparse feature matrix contains the filtered important feature values, reducing the dimension and redundancy of the original feature matrix while maximizing the retention of key signal features. This design not only optimizes transmission efficiency, but also ensures high restoration of compressed features in the decoding and reconstruction process, meeting the real-time and accuracy requirements of data transmission.

[0139] The feature encoding matrix generation step solves the security and integrity problem after data feature extraction, improving the reliability of data transmission. The sparse feature extraction formula is used to extract the sparse matrix from the original signal matrix. Then, the feature matrix is encrypted and integrity checked by the encryption encoding function. The check function ensures that the data has not been tampered with during transmission, thereby enhancing the security and integrity of the data. Through the network state monitoring step, the adaptability problem of dynamic network transmission is solved, ensuring the efficiency and stability of data transmission. Real-time monitoring of network delay, bandwidth and packet loss rate can obtain the state information of the transmission channel in time, providing the basis for subsequent adaptive adjustment of transmission rate, thereby reducing the problems of data loss and low transmission efficiency caused by network state fluctuations. Through the transmission rate calculation step, the data congestion problem caused by network state fluctuations is solved, and the transmission efficiency is improved. The rate adjustment formula is used to dynamically adjust the transmission rate according to the network delay, bandwidth and packet loss rate. This method ensures that the rate is reduced to reduce data loss when the network is congested, and the rate is increased to improve transmission efficiency when the network is good. Through the data packet generation step, the data organization and efficient packaging problem is solved, and the flexibility of the transmission process is improved. The encoded feature matrix is divided into multiple small data packets according to the maximum data packet length, and the integrity of the last data packet is ensured by filling the check bits. This division mechanism optimizes the adaptability of data in different transmission links, reducing the packet loss or data missing phenomenon. Through the data packaging and encoding step, the error checking problem of data packets in the transmission process is solved, and the accuracy of data transmission is improved. The redundancy check bit function is used to add redundancy check bits to each data packet. This method increases the data redundancy information, so that the receiving end can quickly detect and locate the integrity problem of the data packet, reducing the number of retransmissions caused by data packet errors. Through the data sending and checking step, the error detection and recovery problem in the data transmission process is solved, and the reliability of the system is enhanced. The sender sends data packets at the dynamically adjusted rate, and the receiver uses the integrity check formula to check the data integrity in real time. If the check fails, the data packet is automatically retransmitted. This mechanism effectively reduces the communication interruption problem caused by data packet damage or loss.

[0140] The data receiving and checking step solves the problem of data being tampered with or lost during transmission, improving the integrity and reliability of received data. At the receiving end, each data packet is checked for integrity using a verification formula. If the check bit is 0, it means the data has not been tampered with and is received completely; otherwise, the data packet is immediately requested to be retransmitted. This mechanism effectively reduces the storage and subsequent processing of incorrect data, ensuring the accuracy of the data and the reliability of the transmission link. The data decryption step solves the problem of encrypted features not being directly usable, restoring the operability of the data. Using a decryption formula, combined with the inverse matrix and bias term of the verification function, the encoded features in the data packet are decrypted. This process not only restores the original features of the data, but also preserves the integrity and expressiveness of the features, laying the foundation for subsequent data reconstruction. The data reconstruction and splicing step solves the problem of incomplete feature matrix caused by data packet fragmentation, restoring the original structure of the overall feature matrix. Using splicing operations, all decrypted data packets are spliced into a complete feature matrix in order. This process eliminates the discontinuity problem caused by data fragmentation, ensuring the consistency and integrity of the decrypted data, providing accurate input for subsequent data analysis. The data analysis and reconstruction step solves the problem of encoded data features being difficult to directly analyze, restoring the feature information before sparsification. Through the inverse reconstruction function, the decrypted feature matrix is converted back to the sparse feature matrix. This inverse process restores the original expression form of the features, ensuring the sparsity and accuracy of the data, thereby providing high-quality basic data for subsequent data analysis and anomaly detection.

[0141] The input variable definition step solves the problem of unknown variable sources in the reverse reconstruction process, clearly defining the scope and meaning of the required input data. During the input stage, the reverse reconstruction function clearly requires the input of the encoded data matrix, the verification function used in the encoding process, and the data packet. This design ensures that the source of each variable is clear, providing a clear basis for subsequent reverse calculations, effectively reducing the uncertainty of misoperation and calculation process. The output variable definition step solves the problem of ambiguous goals in the reconstruction process, clearly defining the output result of the reverse reconstruction. The output variable is the feature matrix after reverse reconstruction, specifically used to restore the integrity of the original signal features. This step sets a clear goal for the entire reverse reconstruction process, effectively ensuring the coherence of the data processing flow and the practicality of the output. The data decompression step solves the problem of complex data structure caused by encoding compression, achieving partial restoration of feature data. Using the inverse transformation matrix, the encoded data is decompressed to restore the F compressedThe step recovers part of the original structure of the feature data through reverse mapping, reduces the complexity introduced by the compression process, and provides complete basic data for subsequent decryption operations. Through the data decryption step, the problem of encrypted data being unable to be directly used is solved, and the usability of the feature matrix is restored. By decrypting F compressed using the decryption algorithm, F decrypted is generated. This process uses a decryption formula to restore the true information of the feature matrix, effectively solving the problem of encrypted data opacity and ensuring data security and operability. Through the data reverse check step, the consistency problem of data check and recovery during transmission is solved, improving the integrity and reliability of the reconstructed data. By checking the data using the reverse check function, it is ensured that the reconstructed data matrix meets the integrity and consistency requirements of the original feature matrix. This step finally verifies the integrity of the reconstructed data, effectively reducing errors caused by data loss or tampering, and providing a reliable basis for subsequent data analysis.

[0142] By using the adaptive transformation method to construct the feature space of the data, the problem of difficult expression of abnormal features caused by the complexity of data distribution is solved, and flexible representation of data features is realized. The adaptive transformation matrix adjusts according to the dynamic characteristics of the data, so that the transformed feature matrix can better represent the intrinsic characteristics of the data. This step improves the separation ability of the data in the feature space, providing a more accurate basis for subsequent anomaly detection. By using weighted local distance measurement, the problem of ignoring local structure in traditional distance measurement methods in anomaly detection is solved, and the detection accuracy is improved. This step uses adaptive weighting coefficients to emphasize local importance, effectively capturing the abnormal characteristics of data points in the local neighborhood. Compared with global distance measurement, weighted local distance is more sensitive to local changes in data, and can accurately evaluate the abnormality of data points. By setting a dynamic threshold, the problem of static threshold being difficult to adapt to dynamic data distribution is solved, and the flexibility of anomaly detection is improved. The dynamic threshold is calculated based on the mean and standard deviation of the weighted local distance, and the sensitivity is adjusted by a hyperparameter. This design enables the detection system to automatically adjust the judgment standard according to the data distribution, enhancing the adaptability. Through the use of a label vector, the problem of difficulty in automatically labeling abnormal data is solved, and the classification of data points is realized. The label vector is used to divide data points into normal points and abnormal points, directly giving the result of data anomaly detection. This step is fast and efficient, facilitating subsequent data repair and further analysis. Through the repair method based on local patterns, the problem of data integrity decline caused by direct deletion of abnormal data is solved, and the effectiveness of the data is restored. The repair method uses the neighborhood normal data points of the abnormal points and their weighted coefficients to reconstruct the abnormal data. This step not only preserves the overall structure of the data, but also reduces the influence of abnormal data on the analysis results, improving the accuracy of data repair. BRIEF DESCRIPTION OF DRAWINGS

[0143] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0144] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present application.

[0145] Embodiment, refer to Figure 1 , the electric energy metering information transmission method comprises:

[0146] S1, data acquisition and pretreatment;

[0147] The voltage signal is measured by using a voltage measurement module, denoted as V(t);

[0148] The current signal is measured by using a current sensor, denoted as I(t);

[0149] The real-time power signal is calculated by using a power calculation module, denoted as P(t);

[0150] Wherein, t is a time stamp;

[0151] The voltage signal, the current signal and the power signal are time-sequenced and organized to form a matrix:

[0152] ;

[0153] Wherein, X(t) is a matrix containing the voltage signal, the current signal and the power signal;

[0154] The signal is normalized:

[0155] ;

[0156] Wherein, X ij is the jth value of the ith signal in the matrix X(t); is the mean value of the ith signal; is the standard deviation of the ith signal;

[0157] The expression ability of the signal features is enhanced by using a nonlinear transformation:

[0158] ;

[0159] Wherein, α ij , β ij and g ijDynamic change parameter of signal; tanh() and sin() are hyperbolic tangent and sine functions respectively for nonlinear transformation;

[0160] S2, feature extraction and compression;

[0161] S3, data compression and sparsification;

[0162] S4, data transmission;

[0163] S5, data decoding and reconstruction;

[0164] S6, data analysis and anomaly detection.

[0165] Through data acquisition and preprocessing, the problem of complex power signal and insufficient feature expression ability is solved, effectively improving the expression and analysis ability of the signal. Voltage measurement module, current sensor and power calculation module are used to collect voltage signal, current signal and real-time power signal respectively, and time series organization is formed into a matrix. Then, the matrix is normalized to solve the problem of difficult unified processing of signals in different dimensions and amplitude ranges. At the same time, through nonlinear transformation (using hyperbolic tangent and sine functions combined with dynamic change parameter to enhance the nonlinear feature expression ability of the signal), a high-quality foundation is laid for subsequent feature extraction and analysis.

[0166] Through the steps of feature extraction and compression, data sparsification, the problem of high data dimension and low transmission efficiency is solved, effectively reducing the pressure of data storage and transmission. In the process of feature extraction and compression, important feature information in the power signal is extracted, and redundant information is discarded. Then, through sparsification processing, the dimension and redundancy of the data are further reduced, the volume of the data to be transmitted is reduced, the data transmission efficiency is optimized, and the adaptability of the system to large-scale users is improved.

[0167] Through the data transmission step, the problems of data loss and unstable transmission that may occur in the process of power information transmission are solved, and the reliability of data transmission is improved. The data packet processed by feature extraction and sparsification is combined with dynamic transmission strategy, such as adjusting the frequency and priority of data transmission based on network status, to effectively reduce the risk of network congestion. At the same time, the checking mechanism in the transmission process guarantees the integrity and accuracy of the data.

[0168] Through the data decoding and reconstruction step, the problem of insufficient data feature reconstruction accuracy under high compression rate is solved, ensuring the usability and accuracy of the decoded data. In data decoding and reconstruction, the feature information extracted in the compression process is used to recover the original signal features through inverse reconstruction algorithm. This process combines with the checking mechanism to effectively improve the reconstruction accuracy of the decoded data, providing reliable basic data for subsequent analysis.

[0169] By incorporating data analysis and anomaly detection steps, the system addresses the challenges of detecting outliers and the inefficiency of real-time analysis in electricity metering data, thereby enhancing its intelligence level. Utilizing extracted high-quality signal features, combined with dynamic analysis algorithms for real-time data processing and anomaly detection, the system can quickly identify abnormal conditions in the electricity metering process, such as signal noise and abnormal power fluctuations. This effectively reduces false alarms and missed alarms, and strengthens the system's adaptability to complex electrical environments.

[0170] The feature extraction and compression specifically include:

[0171] Extract the time-domain and frequency-domain features of each signal:

[0172] ;

[0173] Among them, f i Let be the time-domain characteristic of the i-th signal; T is the time length of the signal;

[0174] Principal component analysis was performed on the extracted features:

[0175] ;

[0176] Among them, U, V is the matrix after SVD decomposition; F' is the compressed eigenvector matrix;

[0177] Activate the compressed feature vector:

[0178] ;

[0179] Where F is the eigenvector matrix; tanh() is the hyperbolic tangent function, used to enhance the nonlinear expressive power of the features.

[0180] By extracting the time domain and frequency domain features of each signal, the limitations of single-dimensional features in fully reflecting signal characteristics are addressed, effectively improving the comprehensiveness and accuracy of signal description. During feature extraction, time domain features of the signal are calculated to fully reflect the signal's trend in the time dimension. Meanwhile, frequency domain feature analysis reveals the signal's frequency distribution characteristics. The combination of time domain and frequency domain features not only preserves the signal's temporal dynamic information but also enhances the ability to characterize the signal's periodicity, energy distribution, and other characteristics, laying a foundation for subsequent data compression and analysis. By performing principal component analysis on the extracted features, the problem of high-dimensional feature data redundancy and high computational complexity is addressed, effectively reducing the data dimension and optimizing the computational efficiency. Principal component analysis extracts the main feature components through singular value decomposition, and the compressed feature matrix only retains the most representative part of the signal. This step not only reduces data redundancy but also improves storage and transmission efficiency, while avoiding interference from excessive irrelevant features on the analysis results. By activating the compressed feature vectors, the problem of insufficient signal feature expression is addressed, effectively enhancing the non-linear expression ability of the features. Using the hyperbolic tangent function to activate the feature vector matrix further enhances the non-linear expression ability of the signal. This process highlights the dynamic characteristics and weak features hidden in complex signals, making the signal features more distinguishable and providing a more reliable data foundation for subsequent analysis and anomaly detection.

[0181] The data compression and sparsification specifically includes:

[0182] Sparsifying the compressed features reduces data redundancy and improves transmission efficiency:

[0183] ;

[0184] Where F i is the i-th value in the feature matrix; l is a sparsification control parameter used to adjust the intensity of sparsification, and a larger l value will compress more small features; β is the sparsification exponent, β > 1 to enhance sparsity;

[0185] If the condition is met, it is considered an important feature; otherwise, it is considered an unimportant feature, and the unimportant feature is set to 0:

[0186] ;

[0187] Where F final i is the selected feature value; θ is the threshold value used to determine whether the feature is retained;

[0188] The final sparse feature matrix F final is output, which will be used for transmission.

[0189] By sparse processing of the compressed features, the problem of high data redundancy is solved, and the data transmission efficiency is effectively improved. The sparse processing is performed on the feature matrix through a formula, wherein the sparse control parameter and the sparse index β are used to adjust the sparse intensity and the sparse effect. A larger l value causes smaller features to be compressed more, and β>1 enhances the sparsity. This process significantly reduces irrelevant or redundant feature information in the data, reducing the complexity of subsequent processing and transmission. By threshold-based feature value screening, the problem of insufficient feature screening accuracy is solved, and the quality of the final feature matrix is optimized. The sparse features are screened using a conditional screening formula, wherein q is the threshold value, ensuring that only feature values that significantly contribute to signal expression are retained. Zeroing out unimportant features effectively reduces data volume while retaining important feature information, improving feature matrix transmission efficiency and analysis accuracy. By outputting the final sparse feature matrix, the problem of quality degradation of compressed data in traditional methods is solved, significantly improving the reliability and effectiveness of feature transmission. The final sparse feature matrix contains screened important feature values, reducing the dimensionality and redundancy of the original feature matrix while maximizing the retention of key signal features. This design not only optimizes transmission efficiency but also ensures high restoration of compressed features during decoding and reconstruction, meeting real-time and accuracy requirements for data transmission.

[0190] The data transmission specifically includes:

[0191] S41, generating a feature encoding matrix F encoded ;

[0192] From the original input signal matrix X(t), a sparse matrix F is obtained through a sparse feature extraction algorithm sparse :

[0193] ;

[0194] Wherein, X(t) is a signal matrix; A is a feature basis matrix; is a sparse regularization parameter;

[0195] Encrypt F sparse to generate an encoded feature matrix F encoded :

[0196] ;

[0197] Wherein, is an encryption mapping function; () is a feature verification function to ensure data integrity, specifically:

[0198] ;

[0199] wherein W1 is the weight matrix of the check function; b1 is the bias vector;

[0200] S42, network state monitoring;

[0201] Real-time monitoring of network state, obtaining the current delay , bandwidth B(t) and loss packet P(t);

[0202] S43, transmission rate calculation;

[0203] According to the network state, the data transmission rate R(t) is adjusted adaptively:

[0204] ;

[0205] wherein R min is the minimum transmission rate; is the delay difference of the current network; is the maximum delay allowed by the system; is the delay threshold; α, β and are the parameters for adjusting the transmission rate;

[0206] S44, generating data packets;

[0207] Obtaining the length of the maximum data packet transmitted by the network, denoted as L max ;

[0208] Divide F encoded into multiple data packets:

[0209] ;

[0210] wherein P i is the i-th data packet; i represents the index of the data packet; L max represents the maximum length of each data packet;

[0211] Each column of the data packet corresponds to a feature component, and each row of the data packet corresponds to a signal at a specific time step;

[0212] If the last data packet is not L max in length, fill in the check bit 0;

[0213] S45, data packaging and encoding;

[0214] Divide F encoded into multiple small data packets P i according to the transmission rate R(t), and add a redundant check bit to each data packet;

[0215] wherein is a redundancy check function of the data packet, used for checking the integrity of the data packet, in particular:

[0216] ;

[0217] wherein P i,k is the kth data unit in the data packet P i ; and w is a weight coefficient related to the length of the data packet; mod(•,2) is a modulo 2 operation, used for generating a binary check bit;

[0218] S46, data sending and checking;

[0219] P i is sent at a rate of R(t), and the data integrity is checked in real time at the receiving end:

[0220] ;

[0221] If the checking fails, data retransmission is performed.

[0222] ​The feature coding matrix generation step solves the security and integrity problems of data feature extraction, and improves the reliability of data transmission. The sparse feature extraction formula is used to extract a sparse matrix from the original signal matrix. Then, the feature matrix is encrypted and integrity checked by the encryption coding function. The check function ensures that the data has not been tampered with during transmission, thereby enhancing the security and integrity of the data. The network state monitoring step solves the adaptability problem of dynamic network transmission, ensuring the efficiency and stability of data transmission. Real-time monitoring of network delay, bandwidth and packet loss rate can obtain the state information of the transmission channel in time, providing a basis for subsequent adaptive adjustment of transmission rate, thereby reducing the problems of data loss and low transmission efficiency caused by network state fluctuations. The transmission rate calculation step solves the data congestion problem caused by network state fluctuations and improves the transmission efficiency. The rate adjustment formula is used to dynamically adjust the transmission rate according to the network delay, bandwidth and packet loss rate. This method ensures that the rate is reduced when the network is congested to reduce data loss, and the rate is increased when the network is good to improve transmission efficiency. The data packet generation step solves the data organization and efficient packaging problem, and improves the flexibility of the transmission process. The encoded feature matrix is divided into multiple small data packets according to the maximum data packet length, and the integrity of the last data packet is ensured by filling the check bits. This division mechanism optimizes the adaptability of data in different transmission links, reducing the phenomenon of packet loss or data loss. The data packaging and coding step solves the error checking problem of data packets in the transmission process and improves the accuracy of data transmission. The redundancy check bit function is used to add redundancy check bits to each data packet. This method increases the data redundancy information, so that the receiving end can quickly detect and locate the integrity problem of the data packet, reducing the number of retransmissions caused by data packet errors. The data sending and checking step solves the error detection and recovery problem in the data transmission process and enhances the reliability of the system. The sender sends data packets at the dynamically adjusted rate, and the receiver checks the data integrity in real time using the integrity check formula. If the check fails, the data packet is automatically retransmitted. This mechanism effectively reduces the communication interruption problem caused by data packet damage or loss.

[0223] The data decoding and reconstruction specifically includes:

[0224] S51, data receiving and checking;

[0225] For each received data packet P i , first perform integrity check:

[0226] ;

[0227] Where P i is the i-th received data packet; is the check bit generated by the check function;

[0228] If is 0, it means that the data packet is not tampered and the data is complete; otherwise, the receiving end requests retransmission of the data packet;

[0229] S52, data decryption;

[0230] The received data packet P i is decrypted using the data decryption formula:

[0231] ;

[0232] where F decoded i is the decrypted data matrix; Decrypt() is a decryption function, specifically:

[0233] ;

[0234] where b1 is the bias term; W -1 1 is the inverse matrix of the verification function, which ensures the recovery of the feature in the decryption process.

[0235] S53, data reconstruction and splicing;

[0236] All decrypted data segments F decoded i are spliced to reconstruct the complete feature matrix F decoded :

[0237] ;

[0238] where ∪ represents the splicing operation, which splices the decryption results of all data packets in order to form a complete feature matrix;

[0239] S54, data analysis and reconstruction;

[0240] The original data feature F sparse is restored through reverse transformation:

[0241] ;

[0242] where R() represents the inverse reconstruction function, which restores the decrypted feature matrix to the original sparse feature matrix.

[0243] The data receiving and checking step solves the problem of data tampering or loss during transmission, improving the integrity and reliability of received data. At the receiving end, each data packet is checked for integrity, and a verification formula is used to verify the integrity of the data packet. If the check bit is 0, it means the data has not been tampered with and is received completely; otherwise, immediately request retransmission of the data packet. This mechanism effectively reduces the storage and subsequent processing of error data, ensuring the accuracy of the data and the reliability of the transmission link. Through the data decryption step, the problem of encrypted features being unable to be directly used is solved, and the operability of the data is restored. Using the decryption formula, combined with the inverse matrix and bias term of the check function, the encoded features in the data packet are decrypted. This process not only restores the original features of the data, but also preserves the integrity and expressiveness of the features, thereby laying the foundation for subsequent data reconstruction. Through the data reconstruction and splicing step, the problem of incomplete feature matrix caused by data packet fragmentation is solved, and the original structure of the overall feature matrix is restored. Using splicing operations, all decrypted data packets are spliced into a complete feature matrix in order. This process eliminates the discontinuity problem caused by data fragmentation, ensuring the consistency and integrity of the decrypted data, providing accurate input for subsequent data analysis. Through the data analysis and reconstruction step, the problem of encoded data features being difficult to directly use for analysis is solved, and the feature information before sparsification is restored. Through the inverse reconstruction function, the decrypted feature matrix is converted back to the sparse feature matrix. This inverse process restores the original expression form of the features, ensuring the sparsity and accuracy of the data, thereby providing high-quality basic data for subsequent data analysis and anomaly detection.

[0244] The inverse reconstruction function specifically includes:

[0245] The design of the inverse reconstruction function R() is specifically:

[0246] S541, input variables;

[0247] The input variables include F encoded , () and P i ;

[0248] The F encoded is the encoded data matrix; () is the check function used in the encoding process; P i is the data packet;

[0249] S542, output variables;

[0250] The output variable F reconstructed ;

[0251] The F reconstructed is the inverse reconstruction feature matrix;

[0252] S543, data decompression;

[0253] using the inverse transformation matrix T -1 data decompression, and the decompressed data is denoted as F compressed :

[0254] ;

[0255] S544, data decryption;

[0256] The decompressed data is decrypted by a decryption algorithm, and the decrypted data matrix is denoted as F decrypted :

[0257] ;

[0258] S545, data verification;

[0259] ;

[0260] wherein, is the verification function, F reconstructed is the reconstructed data matrix.

[0261] The input variable definition step solves the problem of unknown variable sources in the reverse reconstruction process, and clearly defines the scope and meaning of the required input data. In the input stage, the reverse reconstruction function clearly requires the input of the encoded data matrix, the verification function used in the encoding process, and the data packet. This design ensures that the source of each variable is clear, providing a clear basis for subsequent reverse calculations and effectively reducing the uncertainty of misoperation and calculation process. The output variable definition step solves the problem of ambiguous targets in the reconstruction process, and clearly defines the output results of the reverse reconstruction. The output variable is the feature matrix after reverse reconstruction, which is specifically used to restore the integrity of the original signal features. This step sets a clear goal for the entire reverse reconstruction process, effectively ensuring the coherence of the data processing flow and the practicality of the output. The data decompression step solves the problem of complex data structure caused by encoding compression, and realizes the partial restoration of feature data. The encoded data is decompressed using the inverse transformation matrix, and the compressed data is restored to F compressed . This step restores part of the original structure of the feature data through inverse mapping, reduces the complexity introduced by the compression process, and provides complete basic data for subsequent decryption operations. The data decryption step solves the problem of encrypted data that cannot be directly used, and restores the usability of the feature matrix. F compressed is decrypted by a decryption algorithm to generate F decryptedThis process uses the decryption formula to restore the true information of the feature matrix, effectively solving the problem of encrypted data opacity, ensuring data security and operability. Through the data reverse check step, the consistency problem of data check and recovery in the transmission process is solved, improving the integrity and reliability of the reconstructed data. Through the reverse check function, the data is checked to ensure that the reconstructed data matrix meets the integrity and consistency requirements of the original feature matrix. This step finally verifies the integrity of the reconstructed data, effectively reducing errors caused by data loss or tampering, providing a reliable foundation for subsequent data analysis.

[0262] The data analysis and anomaly detection specifically includes:

[0263] Using an adaptive transformation method to construct the feature space of the data:

[0264] ;

[0265] Where F transformed is the transformed feature matrix; T is the adaptive transformation matrix;

[0266] Using a weighted locality distance measure to evaluate the degree of abnormality of the data points:

[0267] ;

[0268] Where D(x i ) is the abnormality measure of the i-th data point; x i is the transformed feature matrix of the i-th data point; x j is the transformed feature vector of the j-th data point; is a weighting coefficient, used to measure the local importance between points x i and x j ;

[0269] Set a dynamic threshold T d to judge abnormal data points:

[0270] ;

[0271] Where, is the mean of the weighted locality distance of all data points; is the standard deviation of the weighted locality distance of all data points; α is a hyperparameter that controls the sensitivity of the threshold, which can adjust the strictness of anomaly detection;

[0272] According to the threshold T d , mark the abnormal data points;

[0273] Set a marker vector A, whose elements are 1 or 0, indicating whether the data point is abnormal:

[0274] ;

[0275] where A i is the i-th element in the label vector; 1 represents abnormal; 0 represents normal;

[0276] Repair the abnormal data using the repair method based on local pattern, the abnormal data point x i Repair by the following repair algorithm:

[0277] ;

[0278] where x repaired i is the repaired data point; N i is the domain of the data point x i , containing normal data points similar to x i ; is the weight calculated based on the similarity between data points.

[0279] The adaptive transformation method is used to construct the feature space of data, solving the problem of difficult expression of abnormal features caused by complex data distribution, and realizing flexible representation of data features. The adaptive transformation matrix adjusts according to the dynamic characteristics of data, so that the transformed feature matrix can better represent the inherent characteristics of data. This step improves the separation ability of data in the feature space, providing a more accurate basis for subsequent anomaly detection. Through the weighted locality distance measurement, the problem of ignoring local structure in traditional distance measurement method in anomaly detection is solved, and the detection accuracy is improved. This step uses adaptive weighting coefficients to emphasize local importance, effectively capturing the abnormal characteristics of data points in the local neighborhood. Compared with global distance measurement, the weighted locality distance is more sensitive to local changes of data, and can accurately evaluate the abnormality degree of data points. Through the setting of dynamic threshold, the problem that static threshold is difficult to adapt to dynamic data distribution is solved, and the flexibility of anomaly detection is improved. The dynamic threshold is calculated based on the mean and standard deviation of the weighted locality distance, combined with the super parameter adjustment sensitivity. This design enables the detection system to automatically adjust the judgment standard according to the data distribution, enhancing the adaptability. Through the label vector, the problem of difficult automatic labeling of abnormal data is solved, realizing the rapid classification of data points. Using the label vector to divide the data points into normal points and abnormal points directly gives the result of data anomaly detection. This step is fast and efficient, and is convenient for subsequent data repair and further analysis. Through the repair method based on local mode, the problem of data integrity decline caused by direct deletion of abnormal data is solved, and the effectiveness of data is restored. The repair method uses the neighborhood normal data points of abnormal points and their weighting coefficients to reconstruct abnormal data through local information. This step not only preserves the overall structure of data, but also reduces the influence of abnormal data on analysis results, improving the accuracy of data repair.

[0280] The embodiment also provides a system of the power metering information transmission method, including:

[0281] A voltage measurement module is configured to measure a voltage signal.

[0282] A current sensor is configured to measure a current signal.

[0283] A power calculation module is configured to calculate a real-time power signal.

[0284] A network state monitoring module is configured to monitor a network state in real time, and obtain a current delay, bandwidth and lost packet.

[0285] It is to be noted that, in the present text, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0286] The above description is merely preferred embodiments of the present application, and it should be pointed out that, for those skilled in the art, some improvements and refinements can be made without departing from the technical principles of the present application, and these improvements and refinements should also be considered as the protection scope of the present application.

Claims

1. A method for transmitting electricity metering information, characterized in that, include: S1. Data Acquisition and Preprocessing; S2. Feature extraction and compression; S3, data compression and sparsity; S4, Data Transmission; The data transmission specifically includes: S41. Generate the feature coding matrix ; From the original input signal matrix In this process, a sparse matrix is ​​obtained through a sparse feature extraction algorithm. : ;in, It is a matrix containing voltage signals, current signals, and power signals; It is a sparsity regularization parameter; The eigenvector matrix; To find the one that minimizes the objective function ; It is the square of the L2 norm; It is a norm; It is the characteristic basis matrix; Time is the independent variable; right Perform encryption encoding to generate a feature encoding matrix. : ;in, This is a bitwise XOR operation; in, It is an encryption mapping function; () is the feature verification function, specifically: ； in, It is the weight matrix of the verification function; It is the bias vector; It is the hyperbolic tangent function; S42. Network status monitoring; Real-time monitoring of network status to obtain the current network latency difference. ,bandwidth and lost packets ; S43, Transmission rate calculation; Adaptively adjust the data transmission rate based on network conditions: ； in, Minimum transmission rate; This is the maximum allowable latency of the system; This is the delay threshold; , and Parameters for adjusting the transmission rate; For time Transmission rate at that time; S44. Generate data packets; Get the length of the maximum data packet transmitted over the network, denoted as ; Will Divided into multiple data packets: ； in, It is the first One data packet; Indicates the index of the data packet; Indicates the maximum length of each data packet; From According to length Extract the first The first paragraph as the One data packet; Each column of the data packet corresponds to a feature component, and each row of the data packet corresponds to a signal at a specific time step; If the last data packet is insufficient If the length is less than or equal to 0, then fill in the check bit with 0; S45, Data Packaging and Encoding; Will According to transmission rate Divided into multiple small data packets And add redundant check bits to each data packet. ; in The redundancy check bit function for the data packet is as follows: ； in, For data packets The first in One data unit; This is a weighting coefficient, which is related to the data packet length; This is a modulo-2 operation used to generate a binary check bit. This is the sequence number index of the data unit within the data packet; For data packets The length of the data unit; S46. Data transmission and verification; according to Send at the rate Real-time verification of data integrity at the receiving end: ； If the verification fails, the data will be retransmitted. S5, Data Decoding and Reconstruction; S6. Data Analysis and Anomaly Detection.

2. The method for transmitting electricity metering information according to claim 1, characterized in that, The data acquisition and preprocessing specifically include: The voltage signal is measured using a voltage measurement module, denoted as... ; The current signal is measured using a current sensor and denoted as . ; The real-time power signal is calculated using the power calculation module, denoted as... ; The voltage, current, and power signals are time-series organized to form a matrix: ； Normalize the signal: ； in, For matrix The Middle The signal channel in the first The original values ​​of each time sampling point; For the first The mean of each signal channel across all sampling points; For the first The standard deviation of each signal channel across all sampling points; These are the normalized sample values; For signal channel index; For time sampling index within the same signal channel; Using nonlinear transformations to enhance the expressive power of signal features: ； in, , and These are the dynamic parameters of the signal; These are the enhanced eigenvalues ​​after nonlinear transformation.

3. The method for transmitting electricity metering information according to claim 2, characterized in that, The feature extraction and compression specifically include: Extract the time-domain and frequency-domain features of each signal: ； in, For the first The time-domain characteristics of each signal channel; The duration of the signal; This represents the original number of signal channels; Principal component analysis was performed on the extracted features: ； in, , and for The decomposed matrix; This is the compressed eigenvector matrix; Activate the compressed feature vector: ; This is the activated feature vector matrix.

4. The method for transmitting electricity metering information according to claim 3, characterized in that, The data compression and sparsification specifically include: Sparsification is performed on the compressed features: ； in, For the compressed eigenvector matrix at the th The values ​​on each signal channel; These are sparsification control parameters used to adjust the intensity of sparsification. It is the sparsity index. Used to enhance sparsity; These are the eigenvalues ​​after sparsification and shrinkage; The conditions will be met. If a feature is considered important, it is considered unimportant and its value is set to 0. ； in, These are the final feature values ​​after filtering; This is a threshold used to determine whether a feature is retained. Output the final sparse feature matrix This matrix will be used for transmission.

5. The method for transmitting electricity metering information according to claim 4, characterized in that, The data decoding and reconstruction specifically include: S51. Data reception and verification; For each received data packet First, perform an integrity check: ； like A value of 0 indicates that the data packet has not been tampered with and the data is intact; otherwise, the receiving end requests that the data packet be retransmitted. S52, Data Decryption; Use the data decryption formula to process the received data packet. Decryption: ； in, This is the decrypted data matrix; The decryption function is as follows: ； in, This is the inverse of the weight matrix of the verification function; S53, Data Reconstruction and Stitching; All decrypted data fragments The feature matrices are then stitched together to reconstruct the complete feature matrix. : ； Where ∪ represents the concatenation operation, which concatenates the decryption results of all data packets into a complete feature matrix in sequence; The number of spliced ​​segments; S54, Data Analysis and Restructuring; The original data features are restored through inverse transformation. : ； in, This represents the inverse reconstruction function, which restores the decrypted feature matrix to the original sparse feature matrix.

6. The method for transmitting electricity metering information according to claim 5, characterized in that, The reverse reconstruction function specifically includes: Reverse Reconstruction Function The design is as follows: S541. Input variables; Input variables include , (),and ; S542, Output variables; Output variables ; The This is the feature matrix after reverse reconstruction; S543, Data decompression; Using the inverse transformation matrix Perform decompression of the data; the decompressed data is denoted as... : ； S544, Data Decryption; The decompressed data is decrypted using a decryption algorithm, and the decrypted data matrix is ​​denoted as follows. : ； S545, Data Reverse Verification; 。 7. The method for transmitting electricity metering information according to claim 6, characterized in that, The data analysis and anomaly detection specifically include: Constructing the feature space of the data using an adaptive transformation method: ； in, The transformed feature matrix; It is an adaptive transformation matrix; Use a weighted locality distance metric to assess the degree of outlier in the data points: ;in, The square of the Euclidean norm; The number of feature data points involved in the calculation; in, For the first Anomaly measurement for each data point; For the first Transformation feature matrix of each data point; For the first Transformation feature vector of each data point; The weights are calculated based on the similarity between data points; Index the data points; Index for neighborhood data points; Set a dynamic threshold Used to identify outlier data points: ； in, This is the mean of the weighted local distances for all data points; The standard deviation of the weighted local distance for all data points; Hyperparameters for controlling threshold sensitivity; According to the threshold Mark outlier data points; Let there be a label vector Its element is 1 or 0, indicating whether the data point is abnormal: ； in, For the corresponding number in the tag vector Each element represents a data point; 1 indicates an anomaly; 0 indicates normal. A local pattern-based repair method is used to repair abnormal data points. The following repair algorithm is used for repair: ； in, These are the data points after repair; For data points The neighborhood, including and Similar normal data points.

8. A system employing the power metering information transmission method according to claims 1-7, characterized in that, include: Voltage measurement module: used to measure voltage signals; Current sensor: used to measure current signals; Power calculation module: used to calculate real-time power signals; Network status monitoring module: Monitors network status in real time and obtains current latency, bandwidth and packet loss.

Images