Electricity consumption information acquisition test method and system based on self-organizing message

Abstract

The invention belongs to the field of electricity utilization information acquisition and testing, and particularly relates to an electricity utilization information acquisition and testing method and system based on self-organizing messages. The method comprises the following steps: selecting a matched abnormal message preset template to generate an initial message according to a determined abnormal type; determining a selectable editing operation according to the abnormal link position selected by the user; editing the initial message according to an editing operation selected by a user; the link is a simulated link in a full-link architecture for transmitting the electricity utilization information acquisition message, and a node of the link is a device or a component for transmitting the electricity utilization information acquisition message in the electricity utilization information acquisition system; filling the edited message according to preset abnormal message information, and generating a test message to simulate a power utilization information acquisition message; the preset abnormal message information comprises a mapping relationship between different abnormal types and message contents; and transmitting the test message to the electric energy meter or the acquisition terminal, and correspondingly evaluating the fault-tolerant processing capability of the electric energy meter or the acquisition terminal.

Description

Technical Field

[0001] This invention belongs to the field of electricity consumption information collection and testing, specifically relating to a method and system for electricity consumption information collection and testing based on self-organizing messages. Background Technology

[0002] With the development of smart grids, electricity meters, as crucial devices for collecting electricity consumption information, play a vital role in smart meter readings, data transmission, and fault detection. In traditional electricity meter systems, data collection is often conducted through centralized terminals, which receive the meter readings and transmit them to the main station system. With technological advancements, especially the application of smart grids and the Internet of Things (IoT), electricity meters are increasingly demanding higher frequency and more precise monitoring capabilities, while also facing more complex fault scenarios. For example, electricity meters and terminals may experience various anomalies due to hardware failures, communication failures, or poor data transmission, affecting the accuracy and reliability of electricity consumption information collection.

[0003] Therefore, effectively testing and ensuring that electricity meters and terminals can accurately respond to these anomalies is crucial for guaranteeing the efficient operation of smart grid electricity consumption information collection systems and supporting the stability of users' safe production. Existing electricity meter and terminal testing methods mainly focus on monitoring and calibrating their own performance, such as detecting anomalies in parameters like current, voltage, and power. However, sufficient testing and evaluation are still lacking regarding whether electricity meters and collection terminals can correctly parse and respond to anomaly messages. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for collecting and testing electricity consumption information based on self-organizing messages, which solves the problem in the prior art that it is difficult to evaluate whether electricity meters and collection terminals can correctly parse and respond to abnormal messages.

[0005] To achieve the above objectives, the present invention provides a method for collecting and testing electricity consumption information based on self-organizing messages, comprising:

[0006] Based on the determined exception type, select the matching exception message preset template to generate the initial message;

[0007] Based on the link location where the anomaly occurred selected by the user, the available editing operations are determined; based on the editing operations selected by the user, the initial message is edited; the link is a link in the simulated full-link architecture for transmitting electricity information collection messages, and the nodes of the link are devices or components in the electricity information collection system that transmit electricity information collection messages.

[0008] The edited message is filled with the preset abnormal message information to generate a test message to simulate the electricity information collection message corresponding to the abnormal type; the preset abnormal message information contains the mapping relationship between different abnormal types and message content.

[0009] The test message is transmitted to the electricity meter or data acquisition terminal, and the fault tolerance capability of the electricity meter or data acquisition terminal is evaluated accordingly.

[0010] Furthermore, methods for determining the anomaly type include: determining the anomaly type based on the predicted status of the acquired electricity meter and data acquisition terminal;

[0011] The methods for obtaining the predicted status of electricity meters and data acquisition terminals include:

[0012] Based on the acquired energy meter and acquisition terminal status data, global feature data and local feature data are obtained; the global feature data contains status data for a complete time series period, and the local feature data contains short time slice data of current or voltage change segments for a complete time series period.

[0013] Calculate the period of the global feature data, divide the global feature data into multiple period segments according to the time series based on the period, and determine the feature mean and standard deviation of each period segment.

[0014] Initialize the ESN model, set the core parameters of the reservoir and randomly generate the reservoir weight matrix; use the feature mean and standard deviation of the corresponding period segment as the input feature vector, input the reservoir of the ESN model to iterate the input feature vector over time steps, and dynamically update the reservoir state.

[0015] A reservoir state matrix is ​​generated using the reservoir states at all time steps. A target output matrix is ​​generated based on historical data. Feature weights are set by the variance of data features at all time steps, and a feature weight matrix is ​​generated accordingly. The output layer weights are obtained through regularized linear regression based on the reservoir state matrix, the target output matrix, and the feature weight matrix. The data features of a single time step include the statistical characteristics, rate of change characteristics, frequency domain characteristics, and behavioral characteristics of the original measured variables at that time step. The statistical characteristics include at least one of the mean, standard deviation, maximum and minimum values, and volatility of current and voltage, respectively. The rate of change characteristics include at least one of the voltage difference and current change rate between previous and subsequent time steps. The frequency domain characteristics include the Fourier transform dominant frequency energy. The behavioral characteristics include at least one of the load change detection results and data reporting stability evaluation indicators. Historical data includes key indicators recorded by the energy meter and acquisition terminal at historical time points, as well as the corresponding equipment real operating status labels at historical time points. Key indicators include at least one of the following under time synchronization: voltage, current, active power, reactive power, frequency, power factor, signal strength, communication error rate, data reporting frequency, and data missing status.

[0016] The global prediction value for the next time step is calculated using the obtained output layer weights, and the global prediction error is obtained based on the global prediction value and the true value of the reservoir state vector for the next time step.

[0017] Construct a differential equation model of the characteristic values ​​of the electricity meter and the acquisition terminal at the current moment. This differential equation model includes the rate of change independent of the initial time and the rate of change related to the initial time. The initial conditions of this differential equation model are obtained through the global prediction value and the global prediction error.

[0018] Based on the local time series extracted from the local feature data, the dynamic rate of change of the energy meter and acquisition terminal feature values ​​at each time step is calculated, and the change amount corresponding to the dynamic rate of change at the current time and the previous time is calculated accordingly; the rate of change independent of the initial time is optimized based on the dynamic rate of change, and the rate of change related to the initial time is optimized based on the dynamic rate of change and the change amount; the feature values ​​of the energy meter and acquisition terminal at the next time step are predicted based on the optimized differential equation model, and the feature values ​​are used as local predicted values;

[0019] The global and local predicted values ​​are merged to generate a comprehensive abnormal state predicted value; the comprehensive abnormal state predicted value is then input into the SVM model to obtain the predicted state of the energy meter and the acquisition terminal.

[0020] Furthermore, the methods for transmitting test messages to the electricity meter or data acquisition terminal include:

[0021] A self-organizing network is established within the test area. Each device used to transmit test messages is initialized as an independent network node. Each node connects to other nodes wirelessly, identifies information about surrounding nodes, and joins the self-organizing network. A unique node ID is assigned to each node in the self-organizing network for data transmission.

[0022] After the self-organizing network is established, the nodes continuously send probe signals to obtain the values ​​of key parameters of link quality between themselves and other nodes, and obtain a comprehensive link quality evaluation result based on the values ​​of each key parameter. The key parameters of link quality include at least one of signal strength, packet loss rate, latency, and bandwidth.

[0023] In the process of transmitting test messages to the electricity meter or data acquisition terminal, the node prioritizes the link with the best comprehensive link quality evaluation result when transmitting test messages.

[0024] Furthermore, methods for assessing the fault tolerance capabilities of electricity meters or data acquisition terminals include:

[0025] Record the reception time of the test message received by the energy meter or data acquisition terminal and the transmission time when the response message is sent back after the test message is processed. Calculate the time difference between the reception time and the transmission time to obtain the response time. If the response time exceeds the set response threshold, it is determined that the response time is unqualified.

[0026] After the energy meter or data acquisition terminal parses the test message, it checks the consistency between the anomaly type corresponding to the identified test message and the original anomaly type corresponding to the test message. If they are consistent, it is considered a correct identification; otherwise, it is considered an incorrect identification, and an incorrect identification test is performed. In the corresponding number of incorrect identification tests, the ratio of the number of correct identifications to the total number of tests is calculated to obtain the identification accuracy rate. If the identification accuracy rate is less than the set identification accuracy rate threshold, the identification accuracy is deemed unqualified.

[0027] Check the consistency between the response of the electricity meter or data acquisition terminal to the test message and the original response corresponding to the test message. If they are consistent, it is a correct response; otherwise, it is an incorrect response, and an incorrect response test is performed. In the corresponding number of incorrect response tests, the ratio of the number of correct responses to the total number of tests is calculated to obtain the response accuracy rate. If the response accuracy rate is less than the set response accuracy rate threshold, the response accuracy is deemed unqualified.

[0028] Furthermore, the methods for obtaining status data from electricity meters and data collection terminals include:

[0029] Real-time acquisition of multi-dimensional status data from electricity meters and acquisition terminals; noise reduction of the time series corresponding to the multi-dimensional status data by applying a moving average filtering algorithm, detection of missing data values ​​by using time series interpolation to complete the data, and standardization of each data feature to obtain preprocessed multi-dimensional status data as the status data of electricity meters and terminals;

[0030] The collected multidimensional status data includes dynamic features, environmental features, and historical features; the dynamic features include current, voltage, power, power factor, and load; the environmental features refer to temperature and humidity; the historical features include load change rate, current and voltage change rate, and historical fault data.

[0031] Furthermore, the methods for generating test messages by filling the edited message with preset abnormal message information include:

[0032] The Drools rules engine is invoked to populate the edited message fields; then, the complete message content is generated using Java's Velocity template engine, with timestamps, node IDs, and sequence numbers appended.

[0033] Furthermore, the predicted states of the electricity meter and data acquisition terminal include normal state, data anomaly, communication anomaly, and equipment status anomaly; data anomaly includes at least one of voltage anomaly, current anomaly, power anomaly, and power factor anomaly; communication anomaly includes at least one of data loss, message error, and communication interruption; equipment status anomaly includes at least one of load overload, equipment failure, and electricity theft.

[0034] The method for determining the anomaly type based on the predicted status of the acquired electricity meter and data acquisition terminal includes: matching and identifying the predicted status of the electricity meter and data acquisition terminal according to a preset mapping rule library between the predicted status of the electricity meter and data acquisition terminal and the anomaly type to obtain the corresponding anomaly type; the mapping rule library defines the logical association between the predicted status of various types of electricity meters and data acquisition terminals and the anomaly type.

[0035] Furthermore, the node includes a master station, a terminal body, a terminal communication module, an energy meter body, and an energy meter communication module;

[0036] The link locations where an anomaly occurs, which users can select, include: between the master station and the terminal body, between the terminal body and the terminal communication module, between the terminal communication module and the electricity meter communication module, and between the electricity meter communication module and the electricity meter body.

[0037] Furthermore, based on the user-selected location of the link where the anomaly occurred, the selectable editing operations include:

[0038] If the user selects the link location where the anomaly occurred between the master station and the terminal, the available editing operations include inserting the corresponding error field into the message.

[0039] If the terminal body and the terminal communication module are selected, the available editing operations include inserting a data conversion exception field into the message;

[0040] If the terminal communication module and the electricity meter communication module are selected, the available editing operations include inserting random noise data into the message, covering part of the payload, or deleting some fields to simulate the operation of an incomplete message scenario.

[0041] If the communication module between the electricity meter and the electricity meter body is selected, the available editing operations include modifying the status field of the message communication module or inserting hardware fault flags to simulate abnormal operations of the internal voltage and current data of the electricity meter.

[0042] The above-described technical solution of this invention provides a novel method for collecting and testing electricity consumption information based on self-organizing messages, the beneficial effects of which include:

[0043] By using preset templates for different anomaly types and providing users with flexible editing options, efficient generation and customization of anomaly messages are achieved, allowing message content to adapt flexibly to various testing needs. The simulation of the entire link architecture provides a realistic and reliable operating scenario for generating anomaly messages, allowing users to select different node links based on testing objectives to accurately simulate specific communication anomalies or equipment failure scenarios. Automated field filling and generation mechanisms improve the efficiency and accuracy of message generation, while the custom editing function provides high flexibility for complex testing. This complete anomaly message generation system significantly improves the testing effectiveness of energy meters and terminals in responding to diverse anomaly scenarios, providing strong support for the operation, maintenance, and system optimization of smart grid equipment.

[0044] The present invention also provides an electricity consumption information collection and testing system based on self-organizing messages, including a processor, wherein the processor stores executable program instructions, which are executed to implement the above-mentioned electricity consumption information collection and testing method based on self-organizing messages.

[0045] The technical solution of the electricity consumption information collection and testing system based on self-organizing messages described above in this invention can achieve the same beneficial effects as the technical solution of the electricity consumption information collection and testing method based on self-organizing messages described above. Attached Figure Description

[0046] Figure 1 This is a flowchart illustrating the method for collecting and testing electricity consumption information based on self-organizing messages, as described in the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0048] Implementation Method of Electricity Consumption Information Collection and Testing Based on Self-Organizing Messages

[0049] This embodiment presents a technical solution for a power consumption information collection and testing method based on self-organizing messages. This method is based on the real-time status of the power meter and terminal, and adopts anomaly message templates that are more closely matched to the actual status, which can more accurately test the ability of the power meter and terminal to respond to these anomalies. At the same time, it supports users to customize and edit the message templates, thereby providing more controllable and accurate test conditions for the testing of power meters and terminals, and improving the accuracy of fault identification.

[0050] Reference Figure 1 The method for collecting and testing electricity consumption information based on self-organizing messages includes:

[0051] Based on the determined anomaly type, select the matching anomaly message preset template (for example, the anomaly type of electricity meter failure corresponds to an anomaly message preset template with fields including abnormal voltage, current peak and timestamp, and the anomaly type of communication interruption corresponds to an anomaly message preset template with fields including link status, packet loss rate and delay, etc.) to generate the initial message.

[0052] Based on the link location where the anomaly occurred selected by the user, the available editing operations are determined; based on the editing operations selected by the user, the initial message is edited; the link is a link in the simulated full-link architecture for transmitting electricity information collection messages, and the nodes of the link are devices or components in the electricity information collection system that transmit electricity information collection messages.

[0053] The edited message is filled with the preset abnormal message information to generate a test message to simulate the electricity information collection message corresponding to the abnormal type; the preset abnormal message information contains the mapping relationship between different abnormal types and message content.

[0054] The test message is transmitted to the electricity meter or data acquisition terminal, and the fault tolerance capability of the electricity meter or data acquisition terminal is evaluated accordingly.

[0055] Therefore, by using preset templates for different anomaly types and providing users with flexible editing options for these messages, efficient generation and customization of anomaly messages are achieved, allowing message content to adapt flexibly to various testing needs. The simulation of the entire link architecture provides a realistic and reliable operating scenario for generating anomaly messages, allowing users to select different node links based on testing objectives to accurately simulate specific communication anomalies or equipment failure scenarios. Automated field filling and generation mechanisms improve the efficiency and accuracy of message generation, while the custom editing function provides high flexibility for complex testing. Through this complete anomaly message generation system, the testing effectiveness of the response capabilities of energy meters and terminals to diverse anomaly scenarios can be significantly improved, providing strong support for the operation, maintenance, and system optimization of smart grid equipment.

[0056] In one specific implementation, the preset abnormal message information not only includes the mapping relationship between different abnormal types and message content, but also includes the processing measures for different abnormal types; that is, the preset abnormal message information can not only map the corresponding message, but also map the corresponding processing measures.

[0057] Furthermore, in this embodiment, the method for determining the anomaly type includes: determining the anomaly type based on the predicted state of the acquired electricity meter and data acquisition terminal. That is, this electricity consumption information acquisition and testing method can use an anomaly message template that better matches the actual state as the base template for generating the final test message, based on the real-time state of the electricity meter and terminal, thus more accurately testing the ability of the electricity meter and terminal to respond to these anomalies.

[0058] Specifically, the methods for obtaining the predicted status of electricity meters and data acquisition terminals include:

[0059] Based on the acquired energy meter and acquisition terminal status data, global feature data and local feature data are obtained; the global feature data contains status data for a complete time series period, and the local feature data contains short time slice data of current or voltage change segments for a complete time series period.

[0060] Calculate the period of the global feature data, divide the global feature data into multiple period segments according to the time series based on the period, and determine the feature mean and standard deviation of each period segment.

[0061] Initialize the ESN model, set the core parameters of the reservoir and randomly generate the reservoir weight matrix; use the feature mean and standard deviation of the corresponding period segment as the input feature vector, input the reservoir of the ESN model to iterate the input feature vector over time steps, and dynamically update the reservoir state.

[0062] A reservoir state matrix is ​​generated using the reservoir states at all time steps. A target output matrix is ​​generated based on historical data. Feature weights are set by the variance of data features at all time steps, and a feature weight matrix is ​​generated accordingly. The output layer weights are obtained through regularized linear regression based on the reservoir state matrix, the target output matrix, and the feature weight matrix. The data features of a single time step include the statistical characteristics, rate of change characteristics, frequency domain characteristics, and behavioral characteristics of the original measured variables at that time step. The statistical characteristics include at least one of the mean, standard deviation, maximum and minimum values, and volatility of current and voltage, respectively. The rate of change characteristics include at least one of the voltage difference and current change rate between previous and subsequent time steps. The frequency domain characteristics include the Fourier transform dominant frequency energy. The behavioral characteristics include at least one of the load change detection results and data reporting stability evaluation indicators. Historical data includes key indicators recorded by the energy meter and acquisition terminal at historical time points, as well as the corresponding equipment real operating status labels at historical time points. Key indicators include at least one of the following under time synchronization: voltage, current, active power, reactive power, frequency, power factor, signal strength, communication error rate, data reporting frequency, and data missing status.

[0063] In one specific embodiment, the method of setting feature weights by the variance of data features at all time steps includes: using the variance of each data feature in the historical periodic sequence (i.e., at all time steps) as the quantitative basis for feature volatility and discriminative power; further mapping the variance values ​​of all data features through a normalization function (such as proportional normalization, softmax, etc.) to generate normalized feature importance coefficients; and then constructing a feature weight matrix P.

[0064] The global prediction value for the next time step is calculated using the obtained output layer weights, and the global prediction error is obtained based on the global prediction value and the true value of the reservoir state vector for the next time step.

[0065] Construct a differential equation model of the characteristic values ​​of the electricity meter and the acquisition terminal at the current moment. This differential equation model includes the rate of change independent of the initial time and the rate of change related to the initial time. The initial conditions of this differential equation model are obtained through the global prediction value and the global prediction error.

[0066] Based on the local time series extracted from the local feature data, the dynamic rate of change of the energy meter and data acquisition terminal feature values ​​at each time step is calculated, and the corresponding changes in the dynamic rate of change at the current time and the previous time are calculated accordingly. Based on the dynamic rate of change, the rate of change independent of the initial time is optimized. Based on the dynamic rate of change and the corresponding changes in the dynamic rate of change at the current time and the previous time, the rate of change related to the initial time is optimized. Based on the optimized differential equation model, the feature values ​​of the energy meter and data acquisition terminal at the next time step are predicted, and these feature values ​​are used as local predicted values.

[0067] Based on the true values ​​of the energy meter and the data acquisition terminal characteristics at the next time step and the local predicted values ​​at the next time step, the local prediction error is obtained; the local prediction error is fed back to the reservoir of the ESN model, and the connection weights of the reservoir are adjusted accordingly.

[0068] The global and local predicted values ​​are merged to generate a comprehensive abnormal state predicted value; the comprehensive abnormal state predicted value is then input into the SVM model to obtain the predicted state of the energy meter and the acquisition terminal.

[0069] By distinguishing between global and local features in the status data of electricity meters and data acquisition terminals, the overall trend and rapid fluctuations of electricity meters can be comprehensively captured, providing rich and multi-dimensional data input for subsequent anomaly detection. Based on this, the combination of periodic analysis and the ESN model effectively captures the long-term trend changes and dynamic behavior of electricity meters and terminals, providing strong support for accurately predicting potential faults in electricity meters and terminals. The differential equation model is used for detailed modeling of local features, optimizing the timeliness and accuracy of predictions, and enabling rapid identification of sudden anomalies and short-term fluctuations in electricity meters and terminals. The fusion of global and local predictions not only enhances the comprehensiveness and accuracy of prediction results but also provides more precise fault identification in practical applications. Using the SVM model for real-time classification of comprehensive abnormal states can accurately determine the real-time status of electricity meters and terminals and generate corresponding anomaly messages.

[0070] In one specific embodiment, the process of obtaining the predicted state of the above-mentioned energy meter and data acquisition terminal is illustrated as follows:

[0071] First, the acquired status data from the electricity meter and acquisition terminal are divided into a global feature group (global feature data) and a local feature group (local feature data). The global feature group contains complete time series periodic data, which is used to capture the overall trend of the data. The local feature group contains short time slice data of rapid current or voltage changes.

[0072] Rapid current changes specifically refer to situations such as sudden current surges (start-up surges) and intermittent current (abnormal load disconnection); rapid voltage changes specifically refer to situations such as instantaneous voltage drops (voltage drop) and repeated oscillations (voltage regulation failure).

[0073] For screening rapid current and voltage changes, the difference between the current and voltage data at each moment and the previous moment can be calculated and divided by the time interval to obtain the rate of change. The rate of change reflects the magnitude of change per unit time and is the basis for judging whether the signal is "rapidly changing". Furthermore, statistical methods can be used to calculate the standard deviation of current and voltage over the global period, and an empirical coefficient (e.g., 1.5 to 3 times the standard deviation) can be set as a judgment threshold. If the rate of change of current or voltage at a certain moment is greater than the threshold, it indicates that the point is in the rapid change range and is marked as potential short-time slice data.

[0074] If the rate of change exceeds the threshold three times consecutively, it indicates a stable and rapid trend. Therefore, the interval where the rate of change exceeds the threshold three times consecutively is used as the center, and several sampling points (e.g., two time points each) are extended before and after it, and this segment is extracted as a complete short time slice. For the extracted complete short time slices, those with only one or two point jumps that subsequently stabilize are excluded. Edge smoothing or Z-score judgment is performed on the change segments of the short time slices to avoid mistakenly identifying noise as short-term anomalies.

[0075] Local feature groups can also include abrupt changes in the second derivative of voltage / current, i.e., rapid changes in the rate of change (jerk feature); and local feature groups can also include data segments that exceed a preset change threshold.

[0076] In addition, short time slices are data segments whose duration is less than the length of the global period (e.g., less than 10% of period A); that is, the duration of each extracted short time slice should not exceed 10% of the global period (e.g., 30 minutes); for example, if a complete period is 30 minutes, then a single short time slice should not exceed 3 minutes, and is usually controlled between tens of seconds and 2 minutes.

[0077] By dividing the data into global feature groups and local feature groups, the long-term trends and short-term fluctuations of electricity meters and terminals can be accurately extracted, comprehensively reflecting the working status of electricity meters and terminals. This grouping method improves the model's ability to perceive anomalies at different time scales and ensures multi-angle capture of potential anomalies.

[0078] Perform periodic analysis on the global feature set, calculate the period A of the data using the autocorrelation function; divide the time series into multiple period segments based on period A, and calculate the feature mean and standard deviation for each period;

[0079] Global dynamic modeling is performed using the ESN model. The ESN model is initialized, the core parameters of the reservoir are set, and the reservoir weight matrix is ​​randomly generated. The reservoir state is initialized as a zero vector. The core parameters include the number of reservoir nodes, sparsity, and spectral radius.

[0080] Therefore, by analyzing periodicity through autocorrelation function and combining it with ESN model for global modeling, we can accurately capture the long-term trend changes of electricity meters and terminals, and provide global dynamic information for anomaly prediction. The dynamics of ESN model and efficient reservoir computing capabilities make the behavioral modeling of electricity meters and terminals more flexible and accurate.

[0081] Then, the periodic mean and standard deviation are used as input feature vectors to the reservoir, and the reservoir state x(t) is dynamically updated by iterating through the input feature vectors over time.

[0082]

[0083] In the formula, α is the state update rate, which is determined according to the dynamic adjustment requirements of the reservoir, and W in Here, is the input weight matrix, and u(t) is the input feature vector at the current time step. It is periodic information, representing the phase of the current time step within the period. A is the data period, W is the reservoir connection weight matrix, and x(t-1) is the reservoir state vector of the previous time step.

[0084] Store the reservoir state of all time steps as a reservoir state matrix X, generate the target output matrix Y based on historical data, set the feature weights by calculating the variance of the data features, and generate the feature weight matrix P based on the feature weights.

[0085] The output layer weights W are calculated using regularized linear regression. out :

[0086] W out =Y·(P·X) T ·((P·X)·(P·X) T +β·I) -1

[0087] In the formula, P is the feature weight matrix, β is the regularization parameter, which is adjusted according to the data complexity, and I is the identity matrix;

[0088] Historical data specifically refers to the data set that has been collected and labeled during system operation. This includes various key indicators recorded by the energy meter and acquisition terminal at historical moments, such as voltage, current, active / reactive power, frequency, power factor, signal strength, communication error rate, data reporting frequency, and data loss status under time synchronization; combined with maintenance records or manual annotations to establish the actual operating status label of the equipment at that moment (e.g., normal, communication interruption, voltage deviation, hardware failure, etc.). After timestamp alignment, missing data completion, and standardization, this historical data corresponding to key indicators is organized chronologically to generate the target output matrix Y. Each row of the target output matrix Y corresponds to the actual operating status label of the equipment at a certain time step, serving as the target supervision signal for training the ESN output weights.

[0089] Data features refer to the input feature vectors that the system collects and extracts at each time step, which can be used for modeling. These specifically include the statistical characteristics of the original measured variables (such as mean, standard deviation, maximum and minimum values, volatility), rate of change characteristics (such as voltage difference between time steps, current rate of change), frequency domain characteristics (such as Fourier transform frequency energy), and behavioral characteristics (such as load change detection results, data reporting stability assessment indicators). These features are automatically extracted and normalized from the original multi-source data stream through a preprocessing module to form the feature vector sequence of the ESN input, which then participates in the generation of the feature weight matrix.

[0090] The global prediction value at the current time step is calculated based on the output layer weights.

[0091]

[0092] In the formula, λ t This is the time decay coefficient, which controls the prediction's dependence on time. x(t) is the reservoir state vector at the current time step. The global prediction error Δ is calculated. ESN :

[0093]

[0094] In the formula, x t+1 It is the actual value at time t+1;

[0095] The local predicted value is calculated using a differential equation model. Specifically, the differential equation model is first defined (i.e., the differential equation model of the characteristic values ​​of the electricity meter and the acquisition terminal at the current moment):

[0096]

[0097] In the formula, x tThese are the current characteristic values ​​of the electricity meter and data acquisition terminal, where t is time, v is the rate of change independent of the initial time, b is the rate of change dependent on the initial time, and f(x) is the value of the data. t (t) is a function of the dynamic state of the energy meter and the data acquisition terminal;

[0098] Initial conditions for generating the differential equation model based on global predictions and errors.

[0099]

[0100] Local time series are extracted from local feature sets, and the dynamic rate of change R at each time step is calculated based on the local time series. t :

[0101]

[0102] In the formula, x t It is the actual value at the current moment, x t-1 It is the true value at the previous moment, and Δt is the interval between adjacent sampling times;

[0103] The change ΔR is calculated based on the current rate of change and the previous rate of change. t ;

[0104] Then, based on the dynamic rate of change and the amount of change, the parameters of the differential equation modulus are optimized:

[0105] v′=R t

[0106]

[0107] In the formula, v′ is the time-independent rate of change, and b′ is the time-dependent rate of change, which is equivalent to optimizing the rate of change independent of the initial time and the rate of change dependent on the initial time, respectively.

[0108] Predict the local state value at the next time step based on the optimized differential equation model.

[0109]

[0110] In the formula, x t v is the true value at the current moment, v′ is the time-independent rate of change, b′ is the time-dependent rate of change, and t is the actual value at the current moment. t It is the current moment, t t+t It's the next moment.

[0111] By extracting local time series data and calculating dynamic rate of change, the differential equation model can effectively capture the rapid changes and sudden failures of the electricity meter in a short period of time. This process optimizes the accuracy of prediction, reduces the interference of global prediction error on local anomalies, and makes local prediction more accurate.

[0112] By fusing global and local state values, a comprehensive abnormal state prediction value for the electricity meter is generated.

[0113]

[0114] In the formula, It is the global prediction value of the ESN model. These are local predictions from the DE model. These are the initial conditions for the DE model. γ is the fusion coefficient, used to adjust the weights of the global predictions and local corrections. It is set through experimental optimization.

[0115] By integrating global and local forecasts and using a weighted average, a comprehensive anomaly forecast is obtained, making the forecast results more comprehensive and accurate. Global forecasts provide trend judgments, while local forecasts correct for the impact of short-term fluctuations. The combination of the two can optimize the final anomaly forecasting effect.

[0116] Finally, the obtained comprehensive abnormal state prediction values ​​are input into the SVM model for state classification to obtain the predicted state of the energy meter and terminal.

[0117] The predicted status of electricity meters and data acquisition terminals includes normal status, data anomaly, communication anomaly, and equipment status anomaly; data anomaly includes at least one of voltage anomaly, current anomaly, power anomaly, and power factor anomaly; communication anomaly includes at least one of data loss, message error, and communication interruption; equipment status anomaly includes at least one of load overload, equipment failure, and electricity theft.

[0118] The method for determining the anomaly type based on the predicted status of the acquired electricity meters and data acquisition terminals includes: matching and identifying the predicted status of the electricity meters and data acquisition terminals according to a preset mapping rule library between the predicted status and anomaly types to obtain the corresponding anomaly type; the aforementioned mapping rule library defines the logical association between the predicted status of various electricity meters and data acquisition terminals and anomaly types (such as communication interruption, voltage anomaly, bit error drift, load mutation, etc.). Specifically, this mapping rule library can take the form of logical discrimination or machine learning models. For example, through logical discrimination or machine learning models (such as SVM), the state vectors corresponding to the predicted status of the electricity meters and data acquisition terminals are classified into anomaly types, and the anomaly types are sorted and confirmed based on the confidence probability output by the model. Finally, the main anomaly type with the highest confidence is selected as the output result, and this result is bound to a timestamp and device number for anomaly record generation, alarm triggering, and subsequent fault diagnosis and maintenance decisions, ensuring that the predicted status is not only used for status classification, but also more effectively supports anomaly tracing and type identification.

[0119] To ensure that poor data quality of the acquired electricity meter and data acquisition terminal status data does not negatively impact the predicted status of the electricity meter and data acquisition terminal, the methods for acquiring the electricity meter and data acquisition terminal status data include:

[0120] Real-time acquisition (can be performed at a set acquisition frequency) of multi-dimensional status data of electricity meters and acquisition terminals; for the time series corresponding to the multi-dimensional status data, a moving average filtering algorithm is applied to remove noise, data null values ​​are detected and filled using time series interpolation methods, and each data feature is standardized to obtain preprocessed multi-dimensional status data as the status data of electricity meters and terminals;

[0121] The collected multidimensional status data includes dynamic features, environmental features, and historical features; dynamic features include current, voltage, power, power factor, and load; environmental features refer to temperature and humidity; historical features include load change rate, current and voltage change rate, and historical fault data.

[0122] This method of acquiring electricity meter and terminal status data firstly enables precise monitoring of their operational status at different time periods by collecting multi-dimensional status data from electricity meters and terminals in real time. Secondly, by applying a moving average filtering algorithm to the time-series data in the dataset for noise reduction, data anomalies caused by instantaneous fluctuations or noise are eliminated, ensuring smoother, more accurate, and reliable data. Simultaneously, missing data points are detected and filled using time-series interpolation methods to ensure data integrity and avoid the impact of data loss on subsequent analysis. Standardization processing transforms features of different dimensions to a unified standard, making subsequent analysis and model building more efficient and accurate, ensuring data integrity and consistency, and laying a solid foundation for subsequent anomaly prediction; that is, by preprocessing the collected multi-dimensional status data, data quality is improved.

[0123] In one specific embodiment, multidimensional state data can be integrated into a dataset for easier processing. For example, multidimensional state data can be integrated into a dataset, and the time series data in the dataset can be denoised by applying a moving average filtering algorithm. Data null values ​​can be detected and filled using a time series interpolation method. Each data feature can be standardized to obtain a preprocessed dataset, which is the acquired electricity meter and terminal state data.

[0124] In this embodiment, the nodes in the simulated full-link architecture specifically include: master station, terminal body, terminal communication module, energy meter body and energy meter communication module;

[0125] Based on this, the user can select the following link locations where anomalies may occur: between the master station and the terminal body, between the terminal body and the terminal communication module, between the terminal communication module and the electricity meter communication module, and between the electricity meter communication module and the electricity meter body. Users can select the link locations where anomalies may occur through a web interface.

[0126] Afterwards, users can edit the initial message through the web interface, such as adding fields (priority, event cause), modifying field values ​​(e.g., adjusting the anomaly magnitude, occurrence time), and deleting irrelevant fields or adjusting the field order. Specifically, the available editing operations are determined based on the link location where the anomaly occurred, as selected by the user:

[0127] If the user selects the link location where the anomaly occurred between the master station and the terminal, the available editing operations include inserting the operation corresponding to the error field into the message to generate a new "master station-terminal anomaly message" and saving it; the specific operations corresponding to the error field include modifying the source address or destination address, adding CRC error flags, etc.

[0128] If the terminal body and the terminal communication module are selected, the available editing operations include inserting a data conversion exception field into the message to generate a new "terminal communication module exception message" and saving it; the operations for inserting a data conversion exception field into the message include modifying the data payload, simulating error parsing, adding a timeout flag to indicate that the module response timeout, etc.

[0129] If the terminal communication module and the electricity meter communication module are selected, the available editing operations include inserting random noise data into the message, covering part of the payload or deleting some fields, simulating the operation of an incomplete message scenario, in order to generate a new "carrier communication abnormal message" and save it.

[0130] If the communication module between the electricity meter and the electricity meter body is selected, the available editing operations include modifying the status field of the message communication module or inserting hardware fault flags, simulating abnormal operations of the internal voltage and current data of the electricity meter, so as to generate a new "internal abnormal message of the electricity meter" and save it.

[0131] After receiving the edited message, the edited message is filled in according to the preset abnormal message information. The methods for generating test messages include:

[0132] The Drools rules engine is invoked to populate the edited message fields; then, the complete message content is generated using Java's Velocity template engine, with timestamps, node IDs, and sequence numbers appended.

[0133] In a preferred embodiment, before calling the Drools rule engine, it is necessary to define the structure of the message object in the Drools rule engine. The fields in the message that have not been filled in or need to be completed after being edited by the user are encapsulated as "facts" in the form of object attributes. Then, specific rule logic is set in the rule file. For example, by judging the message type, node position, anomaly classification and other conditions, the message objects that need to be completed are filtered in the when matching part of the rule. In the then action part, the default value or inference result of the corresponding field is set, such as automatically filling in error code, timestamp, node ID, anomaly level and other content. Finally, the modified object status is submitted through the Drools update() function to realize the automatic field filling process.

[0134] Specifically, Drools defines the mapping relationship between different exception types and message content, as well as the handling measures for different exception types (i.e., preset exception message information), and generates an exception message table, using Java's Velocity to define preset exception message templates.

[0135] Therefore, by combining the Drools rule engine and the Velocity template engine, efficient generation and custom editing of abnormal messages are achieved, allowing message content to flexibly adapt to different testing needs. Based on the simulation of the entire link architecture, users can select different node links according to the test objectives; the selected node links provide users with selectable editing operations, and users can then customize and edit the initial message to accurately simulate specific communication anomalies or device failure scenarios. The automated field filling and generation mechanism for the edited message improves the efficiency and accuracy of message generation, while the custom editing function provides high flexibility for complex tests.

[0136] After receiving the test message, to more realistically simulate the message transmission scenario, the test message can be transmitted to the electricity meter or data acquisition terminal in the following ways:

[0137] A self-organizing network is established within the test area. Each device used to transmit test messages is initialized as an independent network node. Each node connects to other nodes wirelessly, identifies information about surrounding nodes, and joins the self-organizing network. A unique node ID is assigned to each node in the self-organizing network for data transmission.

[0138] By establishing a self-organizing network, the system can flexibly and automatically organize the network according to the actual environment without the need for complex manual settings and maintenance. Each device, as an independent node, can autonomously join the network and participate in data exchange, which improves the scalability and flexibility of the system. Especially in large-scale deployment scenarios, when a new device joins, it can be quickly identified and integrated into the network, avoiding the complexity and bottleneck problems of node management in traditional networks.

[0139] After the self-organizing network is established, the nodes continuously send probe signals to obtain the values ​​of key parameters of link quality between themselves and other nodes, and obtain a comprehensive link quality evaluation result based on the values ​​of each key parameter. The key parameters of link quality include at least one of signal strength, packet loss rate, latency, and bandwidth.

[0140] In the process of transmitting test messages to the electricity meter or data acquisition terminal, the node prioritizes the link with the best comprehensive link quality evaluation result when transmitting test messages.

[0141] In one specific implementation, a self-organizing network is established within the test area using the DL / T698 protocol. After the initial establishment of the self-organizing network, nodes in the network continuously send probe signals to monitor the link quality between themselves and other nodes. Specifically, the data packet loss rate, network latency, and transmission rate of each link (i.e., the key parameters of link quality that need to be calculated) are first calculated, and then a comprehensive link quality evaluation result is obtained based on the values ​​of each key link quality parameter. The method for obtaining the comprehensive link quality evaluation result based on the values ​​of each key link quality parameter is as follows: the values ​​of each key link quality parameter are weighted and summed according to a certain weight to obtain the comprehensive link quality evaluation result. After obtaining the comprehensive link quality evaluation result, to facilitate nodes to quickly select the link with the best quality, the comprehensive link quality evaluation results are sorted in descending order, and nodes prioritize selecting the link with the highest sorted value to transmit test packets.

[0142] Therefore, by employing ad-hoc networking technology and link quality assessment and optimization, the network communication problems that data acquisition systems for electricity meters and terminals may encounter in large-scale deployments and complex environments are resolved. The ad-hoc network built using the DL / T698 protocol not only flexibly responds to dynamic changes in network devices but also ensures efficient and stable data transmission. Simultaneously, the link quality assessment and optimization mechanism effectively improves communication stability and reliability, preventing abnormal situations that do not fall under test conditions from occurring during test message transmission, thus ensuring the real-time performance and accuracy of testing electricity meters and acquisition terminals.

[0143] After receiving a test message containing abnormal conditions (also referred to as an abnormal message), the electricity meter and data acquisition terminal will perform integrity verification and format parsing on the received abnormal message to ensure the accuracy and integrity of the message content. This step can effectively avoid abnormal responses caused by data loss or format errors during communication, and improve the reliability of the electricity meter and terminal in practical applications. Through the verification and parsing process, it can be ensured that the system's processing of each type of abnormal message conforms to the specifications, preventing the electricity meter or data acquisition terminal from mishandling or failing to respond to abnormal situations due to non-standard processing within the message when facing different types of faults.

[0144] Specifically, methods for assessing the fault tolerance capabilities of electricity meters or data acquisition terminals include:

[0145] Record the reception time of the test message received by the energy meter or data acquisition terminal and the transmission time when the response message is sent back after the test message is processed. Calculate the time difference between the reception time and the transmission time to obtain the response time. If the response time exceeds the set response threshold, it is determined that the response time is unqualified.

[0146] Assessing and optimizing response time can significantly improve the real-time performance and response efficiency of the system, especially in high-load or complex network environments. This ensures that the system responds to various fault events in a timely manner, reduces latency and fault handling time, and thus improves the reliability and service quality of the entire electricity information collection system. In addition, timely optimization of hardware and communication protocols helps to improve the adaptability and long-term stability of electricity meters and terminals.

[0147] After the energy meter or data acquisition terminal parses the test message, it checks the consistency between the anomaly type corresponding to the identified test message and the original anomaly type corresponding to the test message. If they are consistent, it is considered a correct identification; otherwise, it is considered an incorrect identification, and an incorrect identification test is performed. In the corresponding number of incorrect identification tests, the ratio of the number of correct identifications to the total number of tests is calculated to obtain the identification accuracy rate. If the identification accuracy rate is less than the set identification accuracy rate threshold, the identification accuracy is deemed unqualified.

[0148] By conducting tests to accurately identify anomaly types, the electricity information collection system that passes the test can promptly determine the specific anomaly type and take corresponding measures when encountering faults in electricity meters or collection terminals, reducing the risk of misidentification and omission. This not only improves the system's fault diagnosis efficiency in practical applications but also provides maintenance personnel with more accurate fault location information, thereby enhancing the overall intelligence level of the electricity information collection system.

[0149] Check the consistency between the response of the electricity meter or data acquisition terminal to the test message and the original response corresponding to the test message. If they are consistent, it is a correct response; otherwise, it is an incorrect response, and an incorrect response test is performed. In the corresponding number of incorrect response tests, the ratio of the number of correct responses to the total number of tests is calculated to obtain the response accuracy rate. If the response accuracy rate is less than the set response accuracy rate threshold, the response accuracy is deemed unqualified.

[0150] Testing and screening electricity consumption data acquisition devices with high response accuracy, or improving the response accuracy of electricity consumption data acquisition devices, can effectively prevent the system from making erroneous operations when faced with abnormal messages, and ensure that every response is a reasonable response to the correct type of abnormality. By optimizing the terminal's anomaly detection algorithm and strengthening the verification logic, the electricity consumption information acquisition system can respond more accurately and promptly to various complex abnormal situations. Ultimately, this improves the overall processing capacity and intelligence level of the electricity consumption information acquisition system, providing feasible support for the construction of future smart grids.

[0151] In a preferred embodiment, the original exception type and the original response of the test message are recorded in the exception message table generated by Drools, and can be obtained directly by reading the data in the exception message table.

[0152] In one embodiment, different optimization measures can be formulated based on the test results:

[0153] If the response time is unacceptable, optimize the hardware and communication protocol; if the identification accuracy is unacceptable, optimize the anomaly detection algorithm of the electricity meter and terminal; if the response accuracy is unacceptable, strengthen the verification algorithm and improve the error handling logic.

[0154] By analyzing the evaluation results and implementing targeted optimizations, we can not only effectively improve the system's reliability and responsiveness, but also ensure its strong adaptability to cope with potential future problems. Optimization measures for different fault types also help extend the equipment's lifespan and improve the applicability and stability of the electricity information collection system in various scenarios.

[0155] The aforementioned threshold values ​​for response, identification accuracy, and response accuracy were all obtained through statistical analysis of historical data. By comprehensively evaluating and optimizing the fault-tolerant processing capabilities of the electricity meter and data acquisition terminal, the reliability, real-time performance, and intelligence level of the electricity information acquisition system, which includes the electricity meter and data acquisition terminal, can be effectively improved in practical applications. Through integrity verification, response time evaluation, and tests of anomaly type identification accuracy and response accuracy, the electricity information acquisition device can be optimized from multiple dimensions. This not only improves the electricity information acquisition system's ability to handle various anomalies but also provides data support and technical assurance for future upgrades and optimizations of the electricity information acquisition system.

[0156] The assessment of the fault tolerance capability of electricity meters or data acquisition terminals also includes: generating a fault detection report based on the test results; specifically, the fault detection report is a comprehensive report that integrates the test results of response time, identification accuracy, and response accuracy;

[0157] In a preferred embodiment, the fault detection report includes basic test information, operational steps of the test process and corresponding test objectives, abnormal message analysis, response capability analysis and optimization measures;

[0158] Basic information includes the test environment, test date, and equipment information;

[0159] The analysis of abnormal messages includes the type of each abnormal message and the test results, for example:

[0160] For electricity meter fault messages, the test result is whether the fault can be correctly identified and reported; for terminal fault messages, the test result is whether the fault can be correctly identified and reported; for electricity theft messages, the test result is whether abnormal current patterns can be accurately identified; for data verification failure messages, the test result is whether the electricity meter and terminal can handle verification errors.

[0161] The response capability analysis includes the response time, response accuracy, and fault tolerance of the energy meter and data acquisition terminal under each test scenario, and analyzes whether they meet expectations.

[0162] By generating fault detection reports, basic information, operating procedures, abnormal message types, and the response capabilities of electricity meters and terminals from each test can be systematically summarized, providing detailed reference materials for subsequent equipment optimization, fault analysis, and system upgrades. The fault detection report not only covers the testing environment and equipment information but also includes detailed anomaly analysis, enabling technicians to quickly identify the root cause of the problem and thus make targeted adjustments or optimizations to the equipment. Therefore, generating comprehensive fault detection reports helps improve system stability and reliability and is a key step in ensuring the long-term stable operation of equipment.

[0163] Furthermore, after generating a fault detection report, the report can be sorted and stored in the database according to timestamps. The stored data can be backed up regularly, and the security and integrity of the stored and backup data can be checked regularly to generate a detection report. A report retention period and a regular cleanup mechanism can be set to automatically archive test reports that exceed the retention period and move them to the long-term storage area.

[0164] By employing a systematic data storage and management solution, combined with timestamp sorting, regular backups, data integrity checks, report retention period control, and long-term storage management, the reliability, security, and storage efficiency of the electricity information collection system are effectively improved. These measures not only optimize the storage management of fault detection reports but also enhance the system's fault tolerance and the availability of subsequent data analysis, providing strong support for the monitoring, fault diagnosis, and maintenance of power equipment.

[0165] Implementation of an Electricity Consumption Information Acquisition and Testing System Based on Self-Organizing Messages

[0166] This embodiment provides a technical solution for an electricity consumption information collection and testing system based on self-organizing messages. The electricity consumption information collection and testing system includes a processor, which stores executable program instructions. The executable program instructions are executed to implement the electricity consumption information collection and testing method based on self-organizing messages in the above-described embodiment of the electricity consumption information collection and testing method based on self-organizing messages.

[0167] Since the working principle and workflow of the power consumption information collection and testing system based on self-organizing messages in this embodiment have been described in detail in the above-described embodiment of the power consumption information collection and testing method based on self-organizing messages, they will not be repeated here.

[0168] It should be understood that the above-described specific embodiments of the present invention are merely illustrative or explanatory of the principles of the present invention, and do not constitute a limitation thereof.

Claims

1. A method for collecting and testing electricity consumption information based on self-organizing messages, characterized in that, include: Based on the determined exception type, select the matching exception message preset template to generate the initial message; Based on the link location where the anomaly occurred selected by the user, the available editing operations are determined; based on the editing operations selected by the user, the initial message is edited; the link is a link in the simulated full-link architecture for transmitting electricity information collection messages, and the nodes of the link are devices or components in the electricity information collection system that transmit electricity information collection messages. The edited message is filled with the preset abnormal message information to generate a test message to simulate the power consumption information collection message corresponding to the abnormal type. The preset exception message information includes a mapping relationship between different exception types and message content; The test message is transmitted to the electricity meter or data acquisition terminal, and the fault tolerance capability of the electricity meter or data acquisition terminal is evaluated accordingly.

2. The electricity consumption information collection and testing method based on self-organizing messages according to claim 1, characterized in that, The methods for determining the type of anomaly include: determining the type of anomaly based on the predicted status of the acquired electricity meter and the data acquisition terminal; The methods for obtaining the predicted status of electricity meters and data acquisition terminals include: Based on the acquired energy meter and acquisition terminal status data, global feature data and local feature data are obtained; the global feature data contains status data for a complete time series period, and the local feature data contains short time slice data of current or voltage change segments for a complete time series period. Calculate the period of the global feature data, divide the global feature data into multiple period segments according to the time series based on the period, and determine the feature mean and standard deviation of each period segment. Initialize the ESN model, set the core parameters of the reservoir and randomly generate the reservoir weight matrix; use the feature mean and standard deviation of the corresponding period segment as the input feature vector, input the reservoir of the ESN model to iterate the input feature vector over time steps, and dynamically update the reservoir state. A reservoir state matrix is ​​generated using the reservoir states at all time steps. A target output matrix is ​​generated based on historical data. Feature weights are set by the variance of data features at all time steps, and a feature weight matrix is ​​generated accordingly. The output layer weights are obtained through regularized linear regression based on the reservoir state matrix, the target output matrix, and the feature weight matrix. The data features of a single time step include the statistical characteristics, rate of change characteristics, frequency domain characteristics, and behavioral characteristics of the original measured variables at that time step. The statistical characteristics include at least one of the mean, standard deviation, maximum and minimum values, and volatility of current and voltage, respectively. The rate of change characteristics include at least one of the voltage difference and current change rate between previous and subsequent time steps. The frequency domain characteristics include the Fourier transform dominant frequency energy. The behavioral characteristics include at least one of the load change detection results and data reporting stability evaluation indicators. Historical data includes key indicators recorded by the energy meter and acquisition terminal at historical time points, as well as the corresponding equipment real operating status labels at historical time points. Key indicators include at least one of the following under time synchronization: voltage, current, active power, reactive power, frequency, power factor, signal strength, communication error rate, data reporting frequency, and data missing status. The global prediction value for the next time step is calculated using the obtained output layer weights, and the global prediction error is obtained based on the global prediction value and the true value of the reservoir state vector for the next time step. Construct a differential equation model of the characteristic values ​​of the electricity meter and the acquisition terminal at the current moment. This differential equation model includes the rate of change independent of the initial time and the rate of change related to the initial time. The initial conditions of this differential equation model are obtained through the global prediction value and the global prediction error. Based on the local time series extracted from the local feature data, the dynamic rate of change of the energy meter and acquisition terminal feature values ​​at each time step is calculated, and the change amount corresponding to the dynamic rate of change at the current time and the previous time is calculated accordingly; the rate of change independent of the initial time is optimized based on the dynamic rate of change, and the rate of change related to the initial time is optimized based on the dynamic rate of change and the change amount; the feature values ​​of the energy meter and acquisition terminal at the next time step are predicted based on the optimized differential equation model, and the feature values ​​are used as local predicted values; The global and local predicted values ​​are merged to generate a comprehensive abnormal state predicted value; the comprehensive abnormal state predicted value is then input into the SVM model to obtain the predicted state of the energy meter and the acquisition terminal.

3. The electricity consumption information collection and testing method based on self-organizing messages according to claim 1 or 2, characterized in that, Methods for transmitting test messages to the energy meter or data acquisition terminal include: A self-organizing network is established within the test area. Each device used to transmit test messages is initialized as an independent network node. Each node connects to other nodes wirelessly, identifies information about surrounding nodes, and joins the self-organizing network. A unique node ID is assigned to each node in the self-organizing network for data transmission. After the self-organizing network is established, the nodes continuously send probe signals to obtain the values ​​of key parameters of link quality between themselves and other nodes, and obtain a comprehensive link quality evaluation result based on the values ​​of each key parameter. The key parameters of link quality include at least one of signal strength, packet loss rate, latency, and bandwidth. In the process of transmitting test messages to the electricity meter or data acquisition terminal, the node prioritizes the link with the best comprehensive link quality evaluation result when transmitting test messages.

4. The method for collecting and testing electricity consumption information based on self-organizing messages according to claim 1 or 2, characterized in that, Methods for assessing the fault tolerance capabilities of electricity meters or data acquisition terminals include: Record the reception time of the test message received by the energy meter or data acquisition terminal and the transmission time when the response message is sent back after the test message is processed. Calculate the time difference between the reception time and the transmission time to obtain the response time. If the response time exceeds the set response threshold, it is determined that the response time is unqualified. After the energy meter or data acquisition terminal parses the test message, it checks the consistency between the anomaly type corresponding to the identified test message and the original anomaly type corresponding to the test message. If they are consistent, it is considered a correct identification; otherwise, it is considered an incorrect identification, and an incorrect identification test is performed. In the corresponding number of incorrect identification tests, the ratio of the number of correct identifications to the total number of tests is calculated to obtain the identification accuracy rate. If the identification accuracy rate is less than the set identification accuracy rate threshold, the identification accuracy is deemed unqualified. Check the consistency between the response of the electricity meter or data acquisition terminal to the test message and the original response corresponding to the test message. If they are consistent, it is a correct response; otherwise, it is an incorrect response, and an incorrect response test is performed. In the corresponding number of incorrect response tests, the ratio of the number of correct responses to the total number of tests is calculated to obtain the response accuracy rate. If the response accuracy rate is less than the set response accuracy rate threshold, the response accuracy is deemed unqualified.

5. The electricity consumption information collection and testing method based on self-organizing messages according to claim 2, characterized in that, The methods for obtaining status data from electricity meters and data acquisition terminals include: Real-time acquisition of multi-dimensional status data from electricity meters and acquisition terminals; noise reduction of the time series corresponding to the multi-dimensional status data by applying a moving average filtering algorithm, detection of missing data values ​​by using time series interpolation to complete the data, and standardization of each data feature to obtain preprocessed multi-dimensional status data as the status data of electricity meters and terminals; The collected multidimensional status data includes dynamic features, environmental features, and historical features; the dynamic features include current, voltage, power, power factor, and load; the environmental features refer to temperature and humidity; the historical features include load change rate, current and voltage change rate, and historical fault data.

6. The method for collecting and testing electricity consumption information based on self-organizing messages according to claim 1 or 2, characterized in that, The methods for generating test messages include: filling in the edited message with preset exception message information. The Drools rules engine is invoked to populate the edited message fields; then, the complete message content is generated using Java's Velocity template engine, with timestamps, node IDs, and sequence numbers appended.

7. The method for collecting and testing electricity consumption information based on self-organizing messages according to claim 1 or 2, characterized in that, The predicted status of electricity meters and data acquisition terminals includes normal status, data anomaly, communication anomaly, and equipment status anomaly; data anomaly includes at least one of voltage anomaly, current anomaly, power anomaly, and power factor anomaly; communication anomaly includes at least one of data loss, message error, and communication interruption; equipment status anomaly includes at least one of load overload, equipment failure, and electricity theft. The method for determining the anomaly type based on the predicted status of the acquired electricity meter and data acquisition terminal includes: matching and identifying the predicted status of the electricity meter and data acquisition terminal according to a preset mapping rule library between the predicted status of the electricity meter and data acquisition terminal and the anomaly type to obtain the corresponding anomaly type; the mapping rule library defines the logical association between the predicted status of various types of electricity meters and data acquisition terminals and the anomaly type.

8. The method for collecting and testing electricity consumption information based on self-organizing messages according to claim 1 or 2, characterized in that, The node includes a master station, a terminal body, a terminal communication module, an energy meter body, and an energy meter communication module; The link locations where an anomaly occurs, which users can select, include: between the master station and the terminal body, between the terminal body and the terminal communication module, between the terminal communication module and the electricity meter communication module, and between the electricity meter communication module and the electricity meter body.

9. The electricity consumption information collection and testing method based on self-organizing messages according to claim 8, characterized in that, Based on the user-selected location of the link where the anomaly occurred, the available editing operations include: If the user selects the link location where the anomaly occurred between the master station and the terminal, the available editing operations include inserting the corresponding error field into the message. If the terminal body and the terminal communication module are selected, the available editing operations include inserting a data conversion exception field into the message; If the terminal communication module and the electricity meter communication module are selected, the available editing operations include inserting random noise data into the message, covering part of the payload, or deleting some fields to simulate the operation of an incomplete message scenario. If the communication module between the electricity meter and the electricity meter body is selected, the available editing operations include modifying the status field of the message communication module or inserting hardware fault flags to simulate abnormal operations of the internal voltage and current data of the electricity meter.

10. A power consumption information collection and testing system based on self-organizing messages, comprising a processor, wherein the processor stores executable program instructions, characterized in that, The executable program instructions are executed to implement the electricity consumption information collection and testing method based on self-organizing messages as described in any one of claims 1-9.

Images