A multi-dimensional comparison-based electric energy meter operation state recognition method

Abstract

The application discloses a kind of based on multi-dimension comparison electric energy meter operating state identification method, including three redundant steps of current loop, space voltage, time electric quantity, and vector fitting step, comparison analysis step.Each redundant step generates corresponding error redundancy component by exclusive comparison strategy, vector fitting step is formed state fitting vector by confidence weighting and cross-dimension check, and comparison analysis step is mapped state category by the classification model completed by training and outputs diagnostic information.The application realizes multi-dimension parallel check, solves the limitation of single dimension identification, improves the accuracy of electric energy meter operating state identification, can effectively identify measurement deviation, sampling fault and other abnormalities, provides technical support for power operation and maintenance.

Description

Technical Field

[0001] This invention relates to a method for collecting electricity meter data, and more specifically, to a method for identifying the operating status of an electricity meter based on multi-dimensional comparison. Background Technology

[0002] With the deepening of smart grid construction, electricity meters, as core equipment for billing and distribution monitoring, are crucial for ensuring fair trade and operational safety in the power system through real-time and accurate identification of their operating status. Existing electricity meter status monitoring technologies typically rely on single power balance or simple threshold alarms, making it difficult to accurately identify deeper anomalies such as metering deviations, sampling faults, or covert electricity theft under complex load conditions and changing transformer topologies.

[0003] A method for intelligent fusion and anomaly identification of multi-source electricity meter data, published under patent number CN120257220B, is described. This patent identifies anomalies by collecting raw electrical parameters from multiple meters under a unified time reference, combining this with electrical wiring information to construct a cross-device set, and calculating a consistency index of numerical changes and a response offset feature vector. However, this technical solution primarily focuses on comparing electrical parameter trends between devices, and its accuracy in identifying minor metering deviations caused by changes in loop impedance and harmonic distortion is limited. Furthermore, its residual verification mechanism often suffers from a high false alarm rate for atypical anomalies such as electricity theft when dealing with large-scale distribution areas or variable load patterns due to a lack of time-dimensional load pattern feature matching, and it is difficult to perform multi-dimensional, refined classification and diagnosis of meter faults.

[0004] A method for detecting anomalies in a smart meter, disclosed in patent CN116520236B, utilizes principal component decomposition to reduce the dimensionality of standardized multi-source monitoring data and sets alarm thresholds by establishing a Gaussian distribution model and calculating deviation. However, this technical solution employs a statistical black-box analysis, using physically correlated quantities such as voltage and current as principal components for dimensionality reduction, often losing the logical coupling relationships between the original physical dimensions. In practical applications, if specific physical anomalies such as voltage sampling failures or asynchronous voltage dips occur, the statistical deviation may not effectively reflect the underlying physical causes, making cross-dimensional causal verification impossible. Furthermore, it lacks confidence assessments for different diagnostic dimensions, failing to meet the high-reliability diagnostic requirements of modern power methods.

[0005] The aforementioned problems indicate that existing electricity meter status identification technologies still have significant shortcomings in areas such as in-depth mining of multi-dimensional data, physical-logical consistency verification, and refined diagnosis of complex states. Therefore, this invention provides an electricity meter operating status identification method based on multi-dimensional comparison. By constructing three major evaluation systems—current loop redundancy, spatial voltage redundancy, and temporal energy redundancy—and combining a vector fitting strategy with a cross-dimensional verification mechanism, this method aims to significantly improve the accuracy of electricity meter operating status identification and the completeness of fault classification through deep fusion and logical verification of multi-source information. Summary of the Invention

[0006] In view of this, the purpose of this invention is to provide a method for identifying the operating status of an electricity meter based on multi-dimensional comparison.

[0007] To solve the above-mentioned technical problems, the technical solution of the present invention is: a method for identifying the operating status of an energy meter based on multi-dimensional comparison, comprising: The current loop redundancy step is configured with a current loop comparison strategy, which is used to generate a first error redundancy component based on the topological relationship of the metering loop in which the meter is located. The spatial voltage redundancy step is configured with a spatial voltage comparison strategy, which is used to generate a second error redundancy component based on the power supply node where the meter is located. The time-power redundancy step is configured with a time-power comparison strategy, which is used to generate a third error redundancy component based on the meter's power consumption. The vector fitting step includes a vector fitting strategy, which is used to generate a state fitting vector based on a first error redundancy component, a second error redundancy component, and a third error redundancy component. The comparison analysis step includes a comparison analysis strategy, which is used to obtain and output the energy meter status information based on the state fitting vector.

[0008] Furthermore: the current loop comparison strategy includes a branch proportional error sub-strategy, which is used to calculate the branch proportional error between any two energy meters with a common upstream node. The current loop comparison strategy calculates the relative proportional error group corresponding to each energy meter through the branch proportional error sub-strategy. The relative proportional error group includes several branch proportional errors of the energy meter corresponding to different energy meters. The current loop comparison strategy also includes a first error sub-strategy, which is used to generate a first error redundancy component based on the relative proportional error group. The branch proportional error sub-strategy includes calculating the branch proportional error between two energy meters using a preset loop impedance algorithm. This scheme establishes a lateral comparison mechanism between meters by utilizing loop topology relationships, which can quantify the relative metering deviation between different energy meters at the same node, providing a quantifiable technical means for identifying inconsistencies between meters.

[0009] Furthermore: the spatial voltage comparison strategy includes a voltage comparison error sub-strategy, which is used to calculate the voltage comparison error between energy meters located in the same circuit topology. The spatial voltage comparison strategy calculates the corresponding voltage comparison error group for each energy meter through the voltage comparison sub-strategy. The voltage comparison error group includes several voltage comparison errors of different circuit topologies in which the energy meter is located. The spatial voltage comparison strategy also includes a second error sub-strategy, which is used to generate a second error redundancy component based on the voltage comparison error group. The voltage comparison error sub-strategy includes calculating the piecewise impedance of the circuit topology and using the deviation of the regression coefficients calculated according to the linear regression voltage equation as the voltage comparison error. The branch proportional error sub-strategy calculates the branch proportional error between energy meters with a common upstream node, forming a relative proportional error group for each energy meter, and the first error redundancy component is generated by the first error sub-strategy. This scheme, utilizing piecewise impedance calculation and linear regression voltage equations, can identify methodological deviations in the voltage sampling circuit, providing a quantitative analysis method based on circuit principles for detecting voltage faults and meter malfunctions.

[0010] Furthermore: the time-based power consumption comparison strategy includes a power fitting comparison sub-strategy, which is used to calculate the fitting power consumption error of the energy meter; the time-based power consumption comparison strategy also includes a mapping analysis sub-strategy, which is used to calculate a third error redundancy component based on the fitting power consumption error. The power fitting and comparison sub-strategy includes calculating the fitted power generation within a preset time window based on the instantaneous active power feedback from the energy meter, and then subtracting the fitted power generation from the actual power generation to obtain the fitted power error. A voltage comparison error sub-strategy calculates the voltage comparison error between energy meters in the same circuit topology, forming a voltage comparison error group, and a second error redundancy component is generated by a second error sub-strategy. The power fitting and comparison sub-strategy calculates the fitted power error, and a third error redundancy component is generated by a mapping analysis sub-strategy. This scheme compares the instantaneous power integration result with the actual power consumption, effectively detecting power measurement deviations caused by abnormal metering parameters, pulse constant errors, or sampling faults, providing a time-dimensional self-consistent verification method for identifying internal meter function failures.

[0011] Furthermore, the vector fitting strategy includes a component confidence assessment sub-strategy and a vector synthesis sub-strategy. The component confidence assessment sub-strategy is used to assess the confidence of the first, second, and third error redundancy components respectively, generating corresponding confidence coefficients. The vector synthesis sub-strategy is used to weight the first, second, and third error redundancy components according to the confidence coefficients, and then map the weighted components to a preset multi-dimensional state space to generate the state fitting vector. By assessing the confidence of each error redundancy component and generating confidence coefficients through the component confidence assessment sub-strategy, and then weighting and mapping the components to a multi-dimensional state space through the vector synthesis sub-strategy, this scheme solves the problem of inconsistent dimensions and varying reliability of multi-source error components, enabling the state fitting vector to more realistically reflect the actual operating state of the meter, thus improving the accuracy and robustness of subsequent analysis.

[0012] Furthermore, the current loop comparison strategy further includes a loop impedance stability sub-strategy and a current harmonic distortion ratio sub-strategy. The loop impedance stability sub-strategy monitors the equivalent impedance change of the same energy meter's loop under different load conditions, generating an impedance stability error value by calculating the deviation between the sliding window mean of the impedance and the real-time value. The current harmonic distortion ratio sub-strategy compares the current harmonic distortion rates of different energy meters at the same power supply node, generating a harmonic distortion error value by calculating the difference in harmonic distortion rates between the target energy meter and the reference energy meter. The first error sub-strategy generates a first error redundancy component based on the relative proportional error group, the impedance stability error value, and the harmonic distortion error value. By adding the loop impedance stability sub-strategy and the current harmonic distortion ratio sub-strategy, impedance changes and harmonic distortion differences are monitored respectively, and the first error sub-strategy fuses the relative proportional error group, the impedance stability error value, and the harmonic distortion error value to generate the first error redundancy component. This scheme expands the detection methods for the current loop dimension, enabling the identification of anomalies caused by poor contact, shunt power theft, or sampling channel failure, and enhancing the detection coverage of redundant steps in the current loop.

[0013] Furthermore, the spatial voltage comparison strategy also includes a voltage curve clustering sub-strategy and a voltage sag synchronization sub-strategy. The voltage curve clustering sub-strategy is used to perform unsupervised clustering analysis on the voltage curves of all energy meters within the distribution area, compare the cluster centers with the voltage curves of the target energy meter, and generate a clustering deviation error value based on the degree of deviation. The voltage sag synchronization sub-strategy is used to monitor the response synchronization of each energy meter when a voltage sag event occurs. If the target energy meter does not record a voltage sag event synchronized with other reference meters, a sag missing error value is generated. The second error sub-strategy is also used to generate a second error redundancy component by fusing multi-source information based on the voltage comparison error group, the clustering deviation error value, and the sag missing error value. By adding the voltage curve clustering sub-strategy and the voltage sag synchronization sub-strategy, unsupervised clustering analysis and sag event synchronization monitoring are performed respectively, and the second error sub-strategy fuses the voltage comparison error group, the clustering deviation error value, and the sag missing error value to generate the second error redundancy component. This scheme incorporates big data analysis and transient event characteristics, enabling the discovery of voltage anomaly patterns that are difficult to identify using traditional correlation analysis, thereby improving the detection sensitivity of spatial voltage redundancy steps.

[0014] Furthermore, the time-based power consumption comparison strategy also includes a transformer area power balance sub-strategy and a load curve shape matching sub-strategy. The transformer area power balance sub-strategy is used to perform energy conservation analysis on the power consumption of the main meter and each sub-meter in the transformer area, calculate the line loss and imbalance rate of the transformer area, and generate a power balance error value when the imbalance rate of the branch where the target power meter is located exceeds a preset threshold. The load curve shape matching sub-strategy is used to compare the similarity of the load curve shape of the same power meter in different time windows, calculate the shape matching degree through a dynamic time warping algorithm, and generate a shape change error value when the matching degree is lower than a preset threshold. The mapping analysis sub-strategy is also used to generate a third error redundancy component by fusing multi-source information based on the fitted power consumption error, power balance error value, and shape change error component. By adding the transformer area power balance sub-strategy and the load curve shape matching sub-strategy, transformer area-level energy conservation analysis and load shape similarity calculation are performed respectively, and the mapping analysis sub-strategy fuses the fitted power consumption error, power balance error value, and shape change error value to generate a third error redundancy component. This solution expands the detection scope from a single meter to the distribution area level, and from static power consumption comparison to dynamic load characteristic analysis, significantly enhancing the ability to identify anomalies in time-power redundancy steps.

[0015] Furthermore, the vector fitting strategy also includes a cross-dimensional verification sub-strategy. This sub-strategy detects the logical consistency between the first, second, and third error redundancy components. When current loop anomalies and voltage anomalies occur simultaneously, it determines whether a causal relationship exists between them based on circuit principles. If a causal relationship exists, the state fitting vector is dimensionality-reduced; otherwise, a dimensionality-increasing warning is issued. By detecting the logical consistency between the three error redundancy components through the cross-dimensional verification sub-strategy and determining the causal relationship between anomalies based on circuit principles, the state fitting vector is dimensionality-reduced or dimensionality-increasing. This scheme introduces physical logic constraints, enabling the identification of inherent connections between multi-dimensional anomalies, avoiding misjudgments or information redundancy caused by independent alarms, and improving the accuracy and interpretability of comprehensive diagnosis.

[0016] Furthermore, the comparison and analysis strategy includes a state mapping sub-strategy and a diagnostic output sub-strategy. The state mapping sub-strategy has a preset state classification model, which is trained and generated based on historical state fitting vector samples. This model maps the current state fitting vector to preset state categories, including normal state, metering deviation state, voltage sampling fault state, suspected electricity theft state, and communication anomaly state. The diagnostic output sub-strategy generates corresponding diagnostic information based on the mapped state categories and outputs it to a display terminal or remote monitoring platform. The diagnostic information includes anomaly type, confidence level, and suggested handling measures. By mapping the state fitting vector to preset state categories through the state mapping sub-strategy and generating diagnostic information containing anomaly type, confidence level, and suggested handling measures through the diagnostic output sub-strategy, this solution achieves end-to-end mapping from multi-dimensional error components to specific fault types, meeting the needs of refined diagnosis in actual operation and maintenance scenarios.

[0017] The main technical advantages of this invention are reflected in the following aspects: By setting up current loop redundancy steps, spatial voltage redundancy steps, and time-energy redundancy steps, error redundancy components are generated from three independent dimensions: metering loop topology, power supply node spatial distribution, and energy accumulation over time. These components are then synthesized into a state fitting vector through a vector fitting step, and finally, the state information is output through a comparison and analysis step. This architecture enables multi-dimensional parallel verification of the energy meter's operating status, overcoming the limitations of single-dimensional judgment and providing a data foundation for subsequent refined diagnosis. Attached Figure Description

[0018] Figure 1 This is a block diagram of the overall structure of the energy meter operation status identification method based on multi-dimensional comparison according to the present invention.

[0019] Figure 2 This is a flowchart illustrating the current loop comparison strategy.

[0020] Figure 3 This is a flowchart illustrating the vector fitting strategy.

[0021] Figure 4 This is a flowchart illustrating the comparative analysis strategy. Detailed Implementation

[0022] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, so that the technical solution of the present invention can be more easily understood and mastered.

[0023] This embodiment describes in detail a method for identifying the operating status of an energy meter based on multi-dimensional comparison. This method generates error redundancy components through redundancy verification of three dimensions: current loop, spatial voltage, and time-based energy consumption. After vector fitting and cross-dimensional verification, a state fitting vector is formed. Finally, the accurate identification of the operating status of the energy meter is achieved through comparative analysis. Those skilled in the art can complete the construction and operation of the method according to the steps of this embodiment. The specific implementation process of each step and strategy is as follows.

[0024] The current loop redundancy step generates a first error redundancy component by executing a current loop comparison strategy. This strategy includes a branch proportional error sub-strategy, a loop impedance stability sub-strategy, a current harmonic distortion ratio sub-strategy, and a first error sub-strategy. Each sub-strategy is executed sequentially to complete data fusion. The specific implementation steps are as follows: To implement the branch proportional error sub-strategy, firstly, the metering loop topology of all energy meters within the distribution area is obtained, and the set of associated energy meters with a common upstream node corresponding to each energy meter is determined. Then, the branch proportional error between any two energy meters with a common upstream node is calculated using a preset loop impedance algorithm. For the target energy meter... It is related to the electricity meter Branch ratio error between The calculation formula is: ,in For target electricity meter The equivalent loop impedance of the branch is calculated using the loop impedance algorithm based on the line parameters and measured current and voltage values. For target electricity meter The measured effective value of the current. For associated electricity meters The equivalent loop impedance of the branch in question. For associated electricity meters The measured effective value of the current. Based on the above formula, the branch proportional error between the target energy meter and all associated energy meters is calculated, forming the relative proportional error group corresponding to the energy meter. , The number of associated energy meters for the target energy meter.

[0025] The loop impedance stability sub-strategy is implemented to continuously monitor the equivalent impedance of the loop containing the target energy meter and collect real-time values ​​of the loop equivalent impedance under different load conditions. Set the sliding window duration to Calculate the mean impedance within the sliding window: ,in This represents the number of impedance sampling points within the sliding window. For the first The time point of the second sampling, and then through the formula Calculate the impedance stability error value This value is used to characterize the degree of fluctuation in loop impedance under different loads.

[0026] Implement a current harmonic distortion ratio strategy, and select a normal energy meter with a known operating status as a reference energy meter under the same power supply node. Calculate the target electricity meter separately With reference electricity meter Current harmonic distortion rate, current harmonic distortion rate The calculation formula is: ,in This is the effective value of the fundamental current. For the first The effective value of the second harmonic current, and then through the formula Calculate the harmonic distortion error value , The target energy meter's current harmonic distortion rate. The current harmonic distortion rate is used as a reference for electricity meters.

[0027] The first error sub-strategy is executed, which involves fusing multi-source information from the relative proportional error group, impedance stability error value, and harmonic distortion error value to generate the first error redundancy component. The fusion formula is: ,in , , Here are the weighting coefficients for each error term, satisfying... The weighting coefficients are pre-set based on the topological complexity and load characteristics of the metering circuit in the transformer area. This sub-strategy achieves quantitative integration of multiple abnormal features in the current circuit dimension through multi-source error fusion, enabling the first error redundancy component to fully characterize the degree of abnormal operation of the current circuit and improve the coverage of current circuit abnormality detection.

[0028] The spatial voltage redundancy step generates a second error redundancy component by executing a spatial voltage comparison strategy. This strategy includes a voltage comparison error sub-strategy, a voltage curve clustering sub-strategy, a voltage sag synchronization sub-strategy, and a second error sub-strategy. Each sub-strategy is executed according to the process and completes data fusion. The specific implementation steps are as follows: To implement the voltage comparison error sub-strategy, the structural parameters of the circuit topology where the target energy meter is located are first obtained, and the piecewise impedance of the circuit topology is calculated. Then, the measured voltage values ​​of all energy meters within the same circuit topology were collected. Compared with measured current value Based on Kirchhoff's voltage law, a series of linear regression voltage equations are established. ,in This is a vector of measured voltage values. This is a vector of measured current values. The regression slope coefficient, The regression intercept coefficient, The residual term is used to estimate the regression coefficients of the linear regression equation using the least squares method. , Then calculate the deviation of the regression coefficients, slope deviation. intercept deviation , The reference open-circuit voltage of the circuit topology is used, and the comprehensive deviation of the regression coefficients is taken as the voltage comparison error. Based on the above method, the voltage comparison error of the target energy meter under different circuit topologies is calculated, forming the voltage comparison error group corresponding to the energy meter. , The number of circuit topologies in which the target energy meter is located.

[0029] A voltage curve clustering sub-strategy is implemented to collect voltage monitoring data from all electricity meters within the distribution area at the same time dimension, forming the voltage curve for each electricity meter. The K-means algorithm, an unsupervised clustering algorithm, was used to perform cluster analysis on all voltage curves. First, the number of clusters was set. Based on the pre-determined power supply range and line distribution characteristics of the transformer substation, the similarity between voltage curves is calculated using the Euclidean distance formula. , Given the time sampling length of the voltage curve, the cluster center curves of each cluster are obtained through iterative calculation. Then calculate the voltage curve of the target energy meter. The degree of deviation from the center curve of its respective cluster forms the clustering bias error value. This sub-strategy uses unsupervised clustering to mine common features of voltage curves within a distribution area, which can identify abnormal energy meters that deviate from the normal voltage change pattern, thus improving the sensitivity of voltage anomaly detection.

[0030] A voltage sag synchronization strategy is implemented to monitor voltage sag events across the entire distribution area. The criteria for a voltage sag event are a measured voltage value below 90% of the rated voltage and a duration greater than 10ms. When a voltage sag event is detected within the distribution area, the recorded information from all electricity meters for that event is collected. If the target electricity meter does not record a voltage sag event synchronized with other reference electricity meters, a sag missing error value is set. If the target energy meter records the voltage sag event normally, then a sag missing error value is set. This value is used to characterize the synchronicity of the voltage sampling circuit of the electricity meter in response to transient voltage events.

[0031] The second error sub-strategy is implemented to fuse multi-source information from voltage comparison error groups, clustering bias error values, and sag missing error values ​​to generate a second error redundancy component. The fusion formula is: ,in , , Here are the weighting coefficients for each error term, satisfying... The weighting coefficients are pre-set based on the voltage stability of the transformer area and the complexity of the circuit topology. This sub-strategy achieves full-dimensional anomaly quantification of the spatial voltage dimension from steady-state voltage comparison to transient event response by fusing multi-source voltage-related errors, allowing the second error redundancy component to fully characterize the abnormal operation of the energy meter in the spatial voltage dimension.

[0032] The time-power redundancy step generates a third error redundancy component by executing a time-power comparison strategy. This strategy includes a power fitting comparison sub-strategy, a transformer area power balancing sub-strategy, a load curve shape matching sub-strategy, and a mapping analysis sub-strategy. Each sub-strategy is executed sequentially to complete data fusion. The specific implementation steps are as follows: The power fitting and comparison sub-strategy is implemented to collect the real-time instantaneous active power of the target energy meter. Set the preset time window as The fitted power generation within this time window is calculated by integration. Simultaneously, the actual power generation of the target electricity meter within that time window is collected. The actual power generation is obtained from the meter readings of the electricity meter, and then calculated using the formula. Calculate the fitted electric charge error This sub-strategy achieves self-consistency verification of the time dimension metering data by comparing the instantaneous power integral with the actual power generation, and can effectively identify the power metering deviation caused by abnormal metering parameters.

[0033] Implement the power balancing sub-strategy for the distribution area and collect the total power generation from the distribution area's main meter. Power generation of each sub-meter Calculate the bus loss of the station area Then calculate the line loss rate of the transformer area. Simultaneously calculate the power generation of the branch where the target electricity meter is located. Total power generation of this branch The imbalance rate of this branch is obtained. Set the unbalance rate preset threshold as ,like Then the power balance error value is generated. ,like Then the power balance error value is generated. This sub-strategy extends the detection range from a single meter to the substation branch level, and realizes the detection of power anomalies at the substation level through energy conservation analysis.

[0034] The load curve shape matching sub-strategy is implemented to collect the load curves of the target energy meter within different time windows. The load curve is composed of instantaneous active power. Based on the time series structure, the load curve during the normal operating period is selected as the reference curve. The load curve for the period to be tested is selected as the test curve. The dynamic time warping algorithm is used to calculate the morphological matching degree of two load curves. The dynamic time warping algorithm finds the optimal matching path between the two curves and calculates the cumulative distance on the path as the morphological difference value. Then define the morphological matching degree as ,in The preset maximum morphological difference value is used to set the preset threshold for morphological matching. ,like Then, the morphological mutation error value is generated. ,like Then, the morphological mutation error value is generated. This sub-strategy achieves accurate comparison of load curve shape through dynamic time warping algorithm, and can identify abnormal sudden changes in the load characteristics of electricity meters.

[0035] The mapping analysis sub-strategy is executed to fuse multi-source information from the fitted charge error, charge balance error, and morphological change error to generate a third error redundancy component. The fusion formula is: ,in , , Here are the weighting coefficients for each error term, satisfying... The weighting coefficients are pre-set based on the load characteristics of the distribution area and the accuracy requirements of electricity metering. This sub-strategy achieves comprehensive quantification of abnormal characteristics in the time and electricity dimension by integrating the single metering error in the time dimension with the electricity balance and load pattern error at the distribution area level. This allows the third error redundancy component to accurately characterize the operating status of the electricity meter in the time and electricity dimension.

[0036] The vector fitting step generates a state fitting vector based on the first, second, and third error redundancy components by executing a vector fitting strategy. This strategy includes a component confidence evaluation sub-strategy, a vector synthesis sub-strategy, and a cross-dimensional verification sub-strategy. Each sub-strategy is executed sequentially to complete the generation and optimization of the vector. The specific implementation steps are as follows: The component confidence assessment sub-strategy is executed to assess the confidence of the first, second, and third error redundancy components respectively, generating corresponding confidence coefficients. , , The confidence coefficient ranges from 100 to 100. The closer the value is to 1, the higher the confidence level of the corresponding error redundancy component. Confidence assessment is based on the data source quality, sampling frequency, and calculation model accuracy across various dimensions. Higher confidence coefficients are assigned to error redundancy components with complete data sources, high sampling frequencies, and good model fit. Lower confidence coefficients are assigned to error redundancy components with missing data sources, low sampling frequencies, and poor model fit. The specific values ​​for each confidence coefficient are pre-set based on the monitoring equipment configuration and data quality of the distribution area.

[0037] The vector synthesis sub-strategy is executed, and the three error redundancy components are weighted according to the confidence coefficient. The weighting formula is as follows: , , ,in , , The weighted error redundancy components are then used as three-dimensional features and mapped onto a preset three-dimensional state space. The three coordinate axes of this state space correspond to the weighted error redundancy components in the dimensions of current loop, space voltage, and time charge, respectively. The resulting three-dimensional feature vector is the state fitting vector. This sub-strategy addresses the issues of inconsistent dimensions and varying reliability of multi-source error components by using confidence-weighted calculations, enabling the state fitting vector to more accurately reflect the actual operating state of the electricity meter.

[0038] A cross-dimensional verification sub-strategy is executed to detect the logical consistency among the first, second, and third error redundancy components. Based on circuit principles, the causal relationship between anomalies in different dimensions is determined. If a current loop anomaly and a voltage anomaly are detected simultaneously, i.e. , If both exceed their respective anomaly thresholds, the causal relationship between them needs to be determined: If the abnormal current loop is caused by abnormal line impedance, which in turn leads to abnormal voltage sampling values, indicating a clear causal relationship based on circuit principles, then the state fitting vector is dimensionality-reduced from three dimensions to two dimensions, eliminating redundant features caused by the causal relationship; if the abnormal current loop and the abnormal voltage do not have a clear causal relationship based on circuit principles, indicating two independent anomalies, then the state fitting vector is dimensionality-upgraded for warning purposes, adding an anomaly association feature dimension to the original three-dimensional features to form a four-dimensional warning vector. This sub-strategy introduces physical logic constraints, which can identify the inherent correlation between multi-dimensional anomalies, avoid misjudgments or information redundancy caused by independent alarms, and improve the effectiveness and interpretability of the state fitting vector.

[0039] The comparison analysis step obtains and outputs the energy meter's state information based on the state fitting vector by executing a comparison analysis strategy. This strategy includes a state mapping sub-strategy and a diagnostic output sub-strategy. Each sub-strategy is executed sequentially to complete state identification and diagnostic information output. The specific implementation steps are as follows: The state mapping sub-policy is executed, mapping the current state fitting vector to a preset state category using a pre-defined state classification model. This model is a machine learning-based classification model trained using supervised learning. The model's input training samples are a set of historical state fitting vector samples. , To determine the sample size, each historical state fitting vector sample corresponds to a unique labeled state category. These categories include normal state, metering deviation state, voltage sampling fault state, suspected electricity theft state, and communication anomaly state. The sample set must cover all operating states of the electricity meters within the distribution area to ensure sample diversity and representativeness. The model training process is as follows: First, the sample set is divided into a training set (70%) and a test set (30%). Then, a random forest algorithm is selected as the basic classification algorithm. The feature values ​​of the historical state fitting vectors are input into the algorithm model. The model's classification decision tree parameters are optimized through iterative training until the model's classification accuracy on the test set reaches the preset requirements, completing model training. The output of the trained state classification model is the state category label corresponding to the current state fitting vector. Inputting the cross-dimensionally validated state fitting vector into the model allows it to output the corresponding electricity meter state category.

[0040] The diagnostic output sub-strategy is executed, generating corresponding diagnostic information based on the state category obtained from the state mapping sub-strategy. This diagnostic information includes the anomaly type, confidence level, and suggested remedial measures. The confidence level is obtained from the output of the state classification model and represents the probability value of the model determining the state category. Suggested remedial measures are pre-set based on the fault causes and maintenance specifications for different state categories: If the state category is normal, the diagnostic information is no anomaly, confidence level 1.0, and continuous monitoring is recommended; if the state category is metering deviation, the diagnostic information is abnormal metering deviation, corresponding confidence level, and on-site verification of the electricity meter's metering accuracy is recommended; if the state category is voltage sampling fault, the diagnostic information is voltage sampling circuit fault, corresponding confidence level, and checking the voltage sampling line and sampling steps is recommended; if the state category is suspected electricity theft, the diagnostic information is suspected electricity theft, corresponding confidence level, and on-site investigation of electricity usage is recommended; if the state category is communication anomaly, the diagnostic information is data communication anomaly, corresponding confidence level, and checking the electricity meter's communication steps and communication link is recommended. The generated diagnostic information will be synchronously output to the local display terminal and the remote power monitoring platform, realizing local and remote synchronous early warning of the operating status of the electricity meter. This sub-strategy realizes end-to-end mapping from multi-dimensional error components to specific fault types. The output refined diagnostic information can directly guide on-site operation and maintenance work and meet the needs of actual power operation and maintenance scenarios.

[0041] The various steps and strategies of this invention work together to achieve multi-dimensional parallel verification of the operating status of electricity meters from three independent dimensions: current loop, spatial voltage, and time-based power consumption. Through multi-source information fusion and physical logic verification, the status identification results are more accurate and the fault classification is more complete. This effectively overcomes the limitations of single-dimensional judgment in existing technologies and can accurately identify deep-seated abnormal states of electricity meters, such as metering deviation, sampling failure, or concealed electricity theft, under complex load conditions and changing transformer topologies. This provides technical support for fair trade and operational safety in the electricity sector.

[0042] Of course, the above are just typical examples of the present invention. In addition, the present invention may have many other specific embodiments. All technical solutions formed by equivalent substitution or equivalent transformation fall within the scope of protection claimed by the present invention.

Claims

1. A method for identifying the operating status of an electricity meter based on multi-dimensional comparison, characterized in that: include: The current loop redundancy step is configured with a current loop comparison strategy, which is used to generate a first error redundancy component based on the topological relationship of the metering loop in which the meter is located. The spatial voltage redundancy step is configured with a spatial voltage comparison strategy, which is used to generate a second error redundancy component based on the power supply node where the meter is located. The time-power redundancy step is configured with a time-power comparison strategy, which is used to generate a third error redundancy component based on the meter's power consumption. The vector fitting step includes a vector fitting strategy, which is used to generate a state fitting vector based on a first error redundancy component, a second error redundancy component, and a third error redundancy component. The comparison analysis step includes a comparison analysis strategy, which is used to obtain and output the energy meter status information based on the state fitting vector.

2. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 1, characterized in that: The current loop comparison strategy includes a branch proportional error sub-strategy, which is used to calculate the branch proportional error between any two energy meters with a common upstream node. The current loop comparison strategy calculates the relative proportional error group corresponding to each energy meter through the branch proportional error sub-strategy. The relative proportional error group includes several branch proportional errors of the energy meter corresponding to different energy meters. The current loop comparison strategy also includes a first error sub-strategy, which is used to generate a first error redundancy component based on the relative proportional error group. The branch ratio error sub-strategy includes calculating the branch ratio error between two energy meters using a preset loop impedance algorithm.

3. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 2, characterized in that: The spatial voltage comparison strategy includes a voltage comparison error sub-strategy, which is used to calculate the voltage comparison error between energy meters located in the same circuit topology. The spatial voltage comparison strategy calculates the corresponding voltage comparison error group for each energy meter through the voltage comparison sub-strategy. The voltage comparison error group includes several voltage comparison errors of different circuit topologies in which the energy meter is located. The spatial voltage comparison strategy also includes a second error sub-strategy, which is used to generate a second error redundancy component based on the voltage comparison error group. The voltage comparison error sub-strategy includes calculating the piecewise impedance of the circuit topology and calculating the deviation of the regression coefficients as the voltage comparison error by arranging the voltage equations according to linear regression.

4. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 1, characterized in that: The time-based power consumption comparison strategy includes a power fitting comparison sub-strategy, which is used to calculate the fitting power consumption error of the energy meter. The time-based power consumption comparison strategy also includes a mapping analysis sub-strategy, which is used to calculate a third error redundancy component based on the fitting power consumption error. The power fitting comparison strategy includes calculating the fitted power generation within a preset time window based on the instantaneous active power feedback from the electricity meter, and then calculating the difference between the fitted power generation and the actual power generation to obtain the fitted power error.

5. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 1, characterized in that: The vector fitting strategy includes a component confidence evaluation sub-strategy and a vector synthesis sub-strategy. The component confidence evaluation sub-strategy is used to evaluate the confidence of the first error redundancy component, the second error redundancy component, and the third error redundancy component to generate corresponding confidence coefficients. The vector synthesis sub-strategy is used to weight the first error redundancy component, the second error redundancy component, and the third error redundancy component according to the confidence coefficients, and map the weighted components to a preset multi-dimensional state space to generate the state fitting vector.

6. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 2, characterized in that: The current loop comparison strategy further includes a loop impedance stability sub-strategy and a current harmonic distortion ratio sub-strategy. The loop impedance stability sub-strategy is used to monitor the equivalent impedance change of the loop containing the same energy meter under different load conditions, and generates an impedance stability error value by calculating the deviation between the mean value of the sliding window impedance and the real-time value. The current harmonic distortion ratio sub-strategy is used to compare the current harmonic distortion rate of different energy meters under the same power supply node, and generates a harmonic distortion error value by calculating the difference in harmonic distortion rate between the target energy meter and the reference energy meter. The first error sub-strategy generates a first error redundancy component based on the relative proportional error group, the impedance stability error value, and the harmonic distortion error value.

7. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 3, characterized in that: The spatial voltage comparison strategy further includes a voltage curve clustering sub-strategy and a voltage sag synchronization sub-strategy. The voltage curve clustering sub-strategy is used to perform unsupervised clustering analysis on the voltage curves of all electricity meters in the distribution area, compare the cluster centers with the voltage curves of the target electricity meter, and generate a clustering deviation error value based on the degree of deviation. The voltage sag synchronization sub-strategy is used to monitor the response synchronization of each electricity meter when a voltage sag event occurs. If the target electricity meter does not record a voltage sag event synchronized with other reference meters, a sag missing error value is generated. The second error sub-strategy is also used to generate a second error redundancy component by fusing multi-source information based on the voltage comparison error group, the clustering deviation error value, and the sag missing error value.

8. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 4, characterized in that: The time-based power consumption comparison strategy further includes a transformer area power consumption balance sub-strategy and a load curve shape matching sub-strategy. The transformer area power consumption balance sub-strategy is used to perform energy conservation analysis on the power consumption of the main meter and each sub-meter in the transformer area, calculate the line loss and imbalance rate of the transformer area, and generate a power consumption balance error value when the imbalance rate of the branch where the target power meter is located exceeds a preset threshold. The load curve shape matching sub-strategy is used to compare the similarity of the load curve shape of the same power meter in different time windows, calculate the shape matching degree through a dynamic time warping algorithm, and generate a shape change error value when the matching degree is lower than a preset threshold. The mapping analysis sub-strategy is also used to generate a third error redundancy component by fusing multi-source information based on the fitted power consumption error, power consumption balance error value, and shape change error component.

9. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 1, characterized in that: The vector fitting strategy also includes a cross-dimensional verification sub-strategy; the cross-dimensional verification sub-strategy is used to detect the logical consistency between the first error redundancy component, the second error redundancy component and the third error redundancy component. When the current loop abnormality and the voltage abnormality occur at the same time, it is determined whether there is a causal relationship between the two according to the circuit principle. If there is a causal relationship, the state fitting vector is reduced in dimension; if there is no causal relationship, the dimension is increased and warning is issued.

10. The method for identifying the operating status of an energy meter based on multi-dimensional comparison as described in claim 1, characterized in that: The comparison and analysis strategy includes a state mapping sub-strategy and a diagnostic output sub-strategy. The state mapping sub-strategy has a preset state classification model, which is generated by training based on historical state fitting vector samples. This model is used to map the current state fitting vector to a preset state category, which includes normal state, metering deviation state, voltage sampling fault state, suspected electricity theft state, and communication anomaly state. The diagnostic output sub-strategy is used to generate corresponding diagnostic information based on the mapped state category. The diagnostic information includes anomaly type, confidence level, and suggested handling measures.

Images