Electric energy meter data acquisition and management method and system based on voltage non-contact electricity stealing prevention detection

Abstract

The invention relates to the technical field of electric energy meter data acquisition, and provides an electric energy meter data acquisition and management method and system based on voltage non-contact electricity stealing prevention detection, and the method comprises the steps: carrying out the communication with an electric energy meter, reading key data, such as a freezing event, a load curve and a voltage loss record, in the electric energy meter; forming a preliminary data set containing the real-time waveform of the line and the internal recorded data of the electric energy meter; carrying out harmonic analysis on the collected real-time waveform of the line, and identifying abnormal electricity load characteristics; meanwhile, topology diagnosis is carried out, actual wiring is matched with 96 models, and a specific wiring error type and an electric quantity correction coefficient are output; and the cloud marketing system receives and stores the detection report, and automatically generates or updates an inspection work order. The system comprises a data set generation module, a topology diagnosis module and a report uploading module. According to the method, a full-process digital management closed loop of detection, diagnosis, inspection and filing is formed; the system-level integration significantly improves the anti-electricity-stealing inspection efficiency and the metering abnormity tracing ability.

Description

Technical Field

[0001] This invention relates to the field of electricity meter data acquisition technology, and in particular to an electricity meter data acquisition and management method and system based on voltage-based non-contact anti-theft detection. Background Technology

[0002] Traditional electricity meter data collection requires opening the meter for inspection or breaking the lead seal, which poses high operational risks, low efficiency, and a high risk of legal disputes. Electricity theft detection relies on manual inspection, making it difficult to detect wiring errors in real time, such as the 96 error types in three-phase four-wire systems, or internal events like voltage loss or opening records. Furthermore, it cannot simultaneously complete voltage sampling, comprehensive error analysis, and event record extraction without contact, resulting in delayed anti-theft response and fragmented data management.

[0003] Existing technology 1, application number: CN202311655996.5, discloses a dedicated transformer data acquisition terminal electricity information acquisition system, including an electricity meter data acquisition module, a status quantity acquisition module, a pulse quantity acquisition and testing module, and a data transmission module. The electricity meter data acquisition module obtains electricity consumption data in real time from the power equipment of the dedicated transformer user. The pulse quantity acquisition and testing module can effectively calculate the error during data acquisition. By calculating, the amount of error can be reduced, thereby improving the accuracy of the data. It provides services through encryption service and SSAL service through two different PF_UNIX interfaces. After the system starts, only the security management APP can access ESAM. Security startup mainly includes system security startup and application APP security startup. System security startup is verified by the signature and verification of the system image. Although application APP security startup effectively improves data security and prevents data leakage through the signature and verification of the application APP, the error detection is singular: it only calculates the error through the pulse quantity acquisition and testing module, without combining the current transformer ratio, polarity parameters, and harmonic analysis for comprehensive error assessment, making it difficult to fully reflect the metering accuracy. Lack of topology diagnostics: The system does not cover meter wiring topology matching and power quantity correction, and cannot identify wiring errors such as incorrect phase sequence or reversed transformer polarity. Limited data utilization: Although data security is emphasized, the collected data is not deeply integrated with the cloud management system, making it impossible to automatically generate audit work orders or optimize electricity usage records.

[0004] Prior art two, application number: CN202411252195.9, discloses a method and system for processing electricity meter operation data. The system includes an electricity meter circuit measurement data acquisition module, an electricity meter circuit measurement compensation parameter analysis module, and an electricity meter circuit measurement data processing module. It achieves accurate statistical analysis of the comprehensive compensation amount for the instantaneous power consumption measurement of the circuit under different environmental temperature conditions, power frequency changes, and load current changes by scientifically statistically analyzing the instantaneous power consumption parameters and the superimposed compensation parameters. While this improves the accuracy of electricity meter operation data processing, and scientifically and efficiently analyzes the actual instantaneous power consumption parameters based on the instantaneous power consumption parameters and the superimposed compensation parameters, and feeds them back to the power management platform in real time via the Internet of Things communication network, thus improving the quality and reliability of electricity meter data acquisition, it has limitations in environmental compensation: it only compensates for instantaneous power consumption changes based on temperature, frequency, and load changes, and does not address the detection of non-technical losses such as harmonic interference and abnormal load characteristics. Insufficient data dimensions: Relying on instantaneous circuit parameter statistics without simultaneously analyzing internal data of the electricity meter and real-time waveforms of the line makes it difficult to identify the root cause of metering anomalies. Lack of a closed-loop management system: Although data is fed back through the Internet of Things, an automated process of detection-diagnosis-inspection has not been established, still requiring manual intervention to handle anomalies.

[0005] Current technologies 1 and 2 suffer from limitations such as a single source of error, inaccurate identification of abnormal power consumption and wiring errors, and a lack of a closed loop from on-site diagnosis to management decision-making. Therefore, this invention provides a method and system for data acquisition and management of electricity meters based on non-contact voltage-based anti-theft detection. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method for data acquisition and management of electricity meters based on voltage-based non-contact anti-electricity theft detection, comprising the following steps: Based on the preliminary dataset collected, and combined with the turns ratio and polarity parameters of the current transformer, the comprehensive error of the energy meter and the current transformer is calculated to determine the accuracy; harmonic analysis is performed on the collected real-time waveform of the line to identify abnormal power load characteristics; at the same time, topology diagnosis is performed, matching the actual wiring with 96 models to output specific wiring error types and power correction coefficients.

[0007] Optionally, the high-frequency waveform sequence in the preliminary dataset is used as an independent real power source reference to calculate the theoretical value of the instantaneous power of the line; combined with the given current transformer ratio and polarity parameters, the current data in the high-frequency waveform is calibrated and converted; then the calibrated theoretical value of instantaneous power is compared and integrated point by point with the power value of the load curve read from the internal energy meter in the preliminary dataset; a comprehensive error index characterizing the accuracy of the metering device is generated. By performing discrete spectrum transformation on the high-frequency waveform sequence, the amplitude, phase and time-varying pattern of each harmonic are extracted to form an electricity consumption characteristic spectrum; the current electricity consumption characteristic spectrum is deeply matched and the deviation is calculated with the reference spectrum under the historical normal electricity consumption mode to identify and generate a feature identifier indicating the type of abnormal electricity load. The comprehensive error index, abnormal feature identifier, waveform phase relationship and undervoltage record information in the preliminary dataset are fed into a pre-built topology inference engine. The topology inference engine performs multi-dimensional fitting and probability calculation on the actual error mode, harmonic characteristics, event records and 96 models, and finally outputs the most likely wiring error type and the power correction coefficient for power compensation calculation.

[0008] Optionally, the topology inference engine performs multi-dimensional fitting and probability calculations on the actual error patterns, harmonic features, event records, and 96 models, including the following steps: The core of the topology reasoning engine is a pre-generated and stored knowledge base of wiring errors, containing 96 predefined models of possible wiring errors. Each model describes the physical wiring method and also corresponds to a set of theoretical multi-dimensional features, namely the expected error range, harmonic distortion mode, phase angle offset law, and typical voltage and current event sequence when the error occurs. The topology inference engine performs multi-dimensional fitting calculations, matching the input multi-dimensional diagnostic feature vector with the theoretical feature vectors of all 96 models in the wiring error knowledge base in a synchronous manner across the entire space and multiple attributes; it generates a comprehensive matching score for each model, quantifying the degree of agreement between actual measurement data and the theoretical expectations of the model; Based on the comprehensive matching score, the topology inference engine initiates probability calculation and decision-making; it sorts and probabilizes the matching degree of all models, calculates the posterior probability of each model becoming the actual cause of the failure, and selects the model with the highest posterior probability as the diagnostic conclusion.

[0009] Optionally, the process of generating a comprehensive matching score for each model includes the following steps: The multidimensional diagnostic feature vector and the theoretical expected feature vector of a certain model in the wiring error knowledge base are projected and compared in a high-dimensional feature space composed of multiple dimensions such as error, harmonics, phase and events. Different weight coefficients are assigned to each type of feature attribute according to its diagnostic importance, and the deviation between the measured value and the theoretical value in each dimension is calculated respectively. By inversely synthesizing all weighted deviations, a normalized comprehensive matching score for the model is generated.

[0010] Optionally, the process of calculating the deviation between the measured and theoretical values ​​in each dimension includes the following steps: By learning and training from a massive number of historical correct and incorrect wiring cases, weight coefficients are generated, and an inherent diagnostic significance coefficient is pre-set for each type of feature attribute; Assign the corresponding inherent diagnostic significance coefficient to the feature attribute currently being compared as the basis for weighting; use the difference measurement method based on statistical confidence interval to calculate the absolute difference between the measured value and the theoretical expected value of the feature attribute, and normalize the difference value with the standard deviation of the feature within the historical normal fluctuation range to obtain an original deviation; Multiply the original deviation by the assigned weight coefficient to generate a weighted deviation score for the current attribute dimension.

[0011] Optionally, the process of calculating the absolute difference between the measured value and the theoretical expected value of a feature attribute includes the following steps: A pre-generated confidence interval parameter library is called, and the data is formed by statistical analysis of the monitoring data of various characteristic attributes under massive historical normal wiring conditions. For each type of characteristic attribute, a historical normal fluctuation range is defined, which is a numerical interval centered on the mean and bounded by a number of standard deviations plus or minus, as well as a standard deviation benchmark value used to quantify the normal fluctuation amplitude of the characteristic. Calculate the absolute difference between the measured value and the theoretical expected value; convert the absolute difference proportionally to the corresponding standard deviation benchmark value in the confidence interval parameter library, and assess the significance of the absolute difference relative to the normal fluctuation range; By normalizing based on historical statistical information, an original deviation value representing the strength of relative differences is generated.

[0012] Optionally, the process of proportionally converting the absolute difference to the corresponding standard deviation benchmark value in the confidence interval parameter library includes the following steps: A dynamic benchmark index set for target feature attributes is extracted from a pre-established confidence interval parameter library, containing three core elements: the center point of the historical mean, the boundary value of normal fluctuation, and the magnitude of the standard deviation; the current detected feature is automatically matched with the feature identifier code in the parameter library; The attribute measurements obtained by feature extraction from the measured data stream are compared with the expected values ​​output by the theoretical model using algebraic difference calculation to generate unstandardized raw difference quantities. The standard deviation amplitude in the dynamic benchmark index group is used as a calibration reference. By establishing the difference-benchmark ratio relationship, the raw difference quantities are converted into calibration difference coefficients that can be compared across features. Based on the calibrated difference coefficient, a second correction is performed using the normal fluctuation boundary values ​​recorded in the confidence interval parameter library, combined with the characteristics. When the calibrated difference coefficient exceeds the fluctuation boundary, an amplitude compression mechanism is activated. When it is within the boundary range, a linear preservation mechanism is activated to retain the original proportional relationship. The final output is a normalized deviation index with environmental adaptability.

[0013] Optionally, it also includes using a high-voltage clamp meter to collect real-time voltage and current waveform data on the line non-contactly via an insulating rod; communicating with the electricity meter through an interface to read key data such as freezing events, load curves, and undervoltage records inside the electricity meter, forming a preliminary dataset containing real-time waveforms of the line and data recorded inside the electricity meter.

[0014] Optionally, it also includes the detection terminal uploading preliminary datasets, wiring error types, and power correction coefficients to the cloud marketing system via a 4G network; the cloud marketing system receives and stores the report, forms user power consumption profiles, and automatically generates or updates inspection work orders.

[0015] The present invention provides a data acquisition and management system for electricity meters based on voltage-based non-contact anti-theft detection, comprising: The dataset generation module is used to collect real-time voltage and current waveform data on the line non-contactly through an insulating rod using a high-voltage clamp meter; it communicates with the electricity meter through an interface to read key data such as freezing events, load curves and undervoltage records inside the electricity meter, forming a preliminary dataset containing real-time waveforms of the line and data recorded inside the electricity meter. The topology diagnostic module is used to calculate the comprehensive error of the energy meter and current transformer based on the collected preliminary dataset and the turns ratio and polarity parameters of the current transformer, and to judge the accuracy; to perform harmonic analysis on the collected real-time waveform of the line and identify abnormal power load characteristics; and to perform topology diagnostics, matching the actual wiring with 96 models and outputting the specific wiring error type and power correction coefficient. The report upload module is used to detect terminals to upload preliminary datasets, wiring error types, and electricity correction coefficients to the cloud marketing system via the 4G network; the cloud marketing system receives and stores the report, forms user electricity consumption profiles, and automatically generates or updates inspection work orders.

[0016] This invention establishes a multidimensional dataset containing primary-side dynamic waveforms and secondary-side metering data by simultaneously acquiring real-time waveforms of line voltage / current using a high-voltage clamp meter and internal data on freezing events, load curves, and undervoltage records within the energy meter. This enables joint diagnosis of power grid status monitoring and metering equipment performance. Based on current transformer ratio / polarity parameters and real-time waveform data, a comprehensive error calculation module synchronously verifies the overall metering deviation of the energy meter and transformer, solving the problem of difficult equipment error separation in traditional testing. Combining harmonic analysis and load feature extraction algorithms, it identifies abnormal electricity consumption behaviors such as nonlinear loads and electricity theft from both time-domain or waveform distortion and frequency-domain or harmonic components, exhibiting higher detection sensitivity than single-electricity comparison. Through feature matching of 96 preset wiring models, it automatically identifies topological errors such as incorrect phase sequence connection and transformer reverse polarity, generating electricity correction coefficients to correct metering data under incorrect wiring conditions. Relying on a 4G / USB dual-channel data upload mechanism, it seamlessly integrates on-site testing data, waveforms / diagnostic results, and marketing system work orders, forming a closed-loop digital management process encompassing testing, diagnosis, inspection, and archiving. This system-level integration significantly improves the efficiency of anti-electricity theft investigations and the ability to trace metering anomalies.

[0017] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the data acquisition and management method for electricity meters based on voltage-based non-contact anti-theft detection in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the data acquisition and management method for electricity meters based on voltage-based non-contact anti-theft detection in Embodiment 1 of the present invention. Figure 3 This is a process diagram illustrating the formation of a preliminary dataset containing real-time waveforms of the line and internal recorded data of the energy meter in Embodiment 2 of the present invention; Figure 4 This is a flowchart illustrating the process of outputting specific wiring error types and power correction coefficients in Embodiment 4 of the present invention. Figure 5 This is a flowchart illustrating the process of automatically generating or updating inspection work orders in Embodiment 10 of the present invention. Figure 6 This is a block diagram of the electricity meter data acquisition and management system based on voltage-based non-contact anti-theft detection in Embodiment 11 of the present invention. Detailed Implementation

[0020] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0021] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0022] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application. In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0023] Example 1: As Figure 1 As shown, this embodiment of the invention provides a method for data acquisition and management of electricity meters based on voltage-based non-contact anti-electricity theft detection, comprising the following steps: S100: Uses a high-voltage clamp meter to collect real-time voltage and current waveform data on the line without contact through an insulating rod; communicates with the energy meter through an infrared or RS485 interface to read key data such as freezing events, load curves and undervoltage records inside the energy meter, forming a preliminary dataset containing real-time waveforms of the line and data recorded inside the energy meter. Among them, the electricity meter is a single-phase smart electricity meter or a three-phase smart electricity meter; Parameters of a single-phase smart energy meter: Accuracy class A; Nominal voltage 220V; Operating voltage range 0.8Unom~1.15Unom; Current specifications 0.25-0.5(60)A 0.5-1(100)A; Frequency 50Hz+2.5HZ; Operating temperature: Normal operation -25~55℃, Extreme operation -40~70℃; Relative humidity: Annual average <75%, Monthly average <95%; Power consumption: Voltage line, Non-communication state: <1.5W / 8VA; Communication state: <3W; Current line 1VA; Clock accuracy ≤0.5s / d; Design life greater than 16 years; Dimensions 160mmx112mmx71mm; Net weight approximately 1.0kg; Parameters of a three-phase smart energy meter: Accuracy class: Active power: Class B, Class C; Reactive power: Class 2, Class D; Nominal voltage: 3x57.7 / 100V, 3x100V, 3x220 / 380V; Operating voltage range: 0.8Unom~1.15Unom; Current specifications: Transformer connected type: 0.003-0.015(1.2)A, 0.015-0.075(6)A; Direct connection type: 0.2-0.5(60)A, 0.4-1(100)A; Frequency: 50 Hz +2.5HZ; Normal operating temperature: -25~55℃, extreme operating temperature: -40~70℃; Power consumption: Voltage <1.5W / 6VA, Current <0.2VA (10Itr<10A)<0.4VA (10ltr>10A); Annual average relative humidity <75%, monthly average <95%; Clock accuracy: no more than 0.5s / d; Design life: more than 16 years; Dimensions: 290mm x 170mm x 85mm; Net weight: approximately 2kg; S200: Based on the collected preliminary dataset, combined with the turns ratio and polarity parameters of the current transformer, calculate the comprehensive error of the energy meter and the current transformer, and judge the accuracy; perform harmonic analysis on the collected real-time waveform of the line to identify abnormal power load characteristics; at the same time, perform topology diagnosis, match the actual wiring with 96 models, and output the specific wiring error type and power correction coefficient; S300: The detection terminal uploads the preliminary dataset, wiring error type, and power correction coefficient to the cloud marketing system via the 4G network; the cloud marketing system receives and stores the report, forms a user power consumption profile, and automatically generates or updates inspection work orders.

[0024] The working principle and beneficial effects of the above technical solution are as follows: First, a high-voltage clamp meter is used to collect real-time voltage and current waveform data on the line non-contactly via an insulating rod. Then, through an infrared or RS485 interface, it communicates with the electricity meter to read key data such as freezing events, load curves, and undervoltage records, forming a preliminary dataset containing real-time line waveforms and internal meter data. Second, based on the collected preliminary dataset, combined with the current transformer's ratio and polarity parameters, the comprehensive error of the electricity meter and current transformer is calculated to determine accuracy. Harmonic analysis is performed on the collected real-time line waveforms to identify abnormal power load characteristics. Simultaneously, topology diagnosis is performed, matching the actual wiring with 96 models to output specific wiring error types and power correction coefficients. Finally, the detection terminal uploads the preliminary dataset, wiring error types, and power correction coefficients to the cloud marketing system via a 4G network. The cloud marketing system receives and stores the report, forming user electricity consumption files and automatically generating or updating inspection work orders. (Specific principles are as follows...) Figure 2 The above scheme establishes a multidimensional dataset containing primary-side dynamic waveforms and secondary-side metering data by simultaneously acquiring real-time waveforms of line voltage / current using a high-voltage clamp meter and data on internal freezing events, load curves, and undervoltage records of the energy meter. This enables joint diagnosis of power grid status monitoring and metering equipment performance. Based on the current transformer ratio / polarity parameters and real-time waveform data, the scheme synchronously verifies the overall metering deviation of the energy meter and transformer through a comprehensive error calculation module, solving the problem of difficult equipment error separation in traditional testing. Combining harmonic analysis and load feature extraction algorithms, the scheme identifies abnormal electricity consumption behaviors such as nonlinear loads and electricity theft from both time-domain or waveform distortion and frequency-domain or harmonic components, exhibiting higher detection sensitivity than single-electricity comparison. Through feature matching of 96 preset wiring models, the scheme automatically identifies topological errors such as incorrect phase sequence connection and transformer reverse polarity, and generates electricity correction coefficients to correct metering data under incorrect wiring conditions. Leveraging a dual-channel data upload mechanism via 4G / USB flash drive, on-site testing data, waveforms / diagnostic results are seamlessly integrated with the marketing system's work order flow, forming a closed-loop digital management system encompassing testing, diagnosis, inspection, and archiving. This system-level integration significantly improves the efficiency of anti-electricity theft inspections and the ability to trace metering anomalies.

[0025] Example 2: Figure 3 As shown, based on Embodiment 1, the process for forming a preliminary dataset containing real-time waveforms of the line and internal recorded data of the energy meter, provided by this embodiment of the invention, includes the following steps: S101: Utilizing the principle of high-voltage electromagnetic coupling, the sensing unit at the top of the insulating rod captures information on changes in the electric and magnetic fields around the line; it converts the real-time voltage and current waveform characteristics of the line conductors into an initial analog signal sequence; and it sends instruction codes conforming to the communication protocol to the energy meter through photoelectric conversion or digital serial communication interface. The instruction codes trigger the response of its internal storage unit, retrieving key data blocks such as the freeze event log, load curve array, and undervoltage record list pre-embedded in the firmware. S102: Analog signal sequences and key data blocks are captured in parallel under the scheduling of the time synchronization mechanism. Based on a unified time reference axis, the high-frequency sampling sequence obtained by analog-to-digital conversion of the analog signal sequence is aligned and spliced ​​with the frozen data read from inside the energy meter, the load curve matrix, and the timestamp of the undervoltage event. S103: Generate a preliminary dataset with inherent correlation and unified spatiotemporal coordinates, containing sequences of original voltage and current waveform features and key data records.

[0026] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first utilizes the principle of high-voltage electromagnetic coupling to capture the electric and magnetic field changes in the space surrounding the line through the sensing unit at the top of the insulating rod; the real-time voltage and current waveform characteristics of the line conductor are converted into an initial analog signal sequence; through photoelectric conversion or digital serial communication interface, instruction codes conforming to its communication protocol are sent to the energy meter, and the instruction codes trigger the response of its internal storage unit to retrieve key data blocks such as the freeze event log, load curve array, and undervoltage record list pre-embedded in the firmware; secondly, the analog signal sequence and key data blocks are captured in parallel under the scheduling of the time synchronization mechanism, and according to a unified time reference axis, the high-frequency sampling sequence obtained by analog-to-digital conversion of the analog signal sequence is aligned and spliced ​​with the freeze data, load curve matrix, and undervoltage event timestamp read from inside the energy meter; finally, a preliminary dataset with inherent correlation and unified spatiotemporal coordinates is generated, containing the original voltage and current waveform characteristics sequence and key data records. The above solution breaks through the limitations of traditional single data sources and establishes a joint dataset of line operating conditions and metering equipment responses with millisecond-level time synchronization accuracy. The inherent correlation of its heterogeneous data can be traced back to the original acquisition stage through spatiotemporal coordinates.

[0027] Example 3: Based on Example 2, the process of aligning and stitching the frozen data, load curve matrix, and undervoltage event timestamps read from the internal energy meter provided in this embodiment of the invention includes the following steps: S1021: The analog signal sequence is quantized into a digital high-frequency waveform sampling sequence with an inherently high sampling rate, containing instantaneous voltage and current value information. Each sampling point is assigned a time stamp provided by a clock source, forming a high-resolution waveform image with time as the horizontal axis and sampling value as the vertical axis; key data blocks are parsed in parallel and carry event timestamps assigned by the energy meter clock. S1022: Using the time stamp of the high-frequency waveform sampling sequence as a reference system, the timestamps of various records in the key data block are checked for consistency and recalibrated. Electrical event points with a sudden rise or fall in voltage in the high-frequency waveform sampling sequence are identified and extracted. These points are then mapped and correlated with the occurrence times of the undervoltage event and load change event recorded in the key data block. S1023: After temporal alignment and semantic mapping, the high-frequency sampling sequence and key data blocks are fused into a preliminary temporal-aligned dataset.

[0028] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, the analog signal sequence is first quantized into a digital high-frequency waveform sampling sequence with an inherent high sampling rate, containing instantaneous voltage and current value information. Each sampling point is assigned a time stamp provided by a clock source, forming a high-resolution waveform image with time as the horizontal axis and sampling value as the vertical axis. The key data block is parsed in parallel, carrying the event timestamp assigned by the electricity meter clock. Secondly, using the time stamp of the high-frequency waveform sampling sequence as a reference system, the timestamps of various records in the key data block are checked for consistency and recalibrated. Electrical event points with a sudden rise or fall in voltage at the start time in the high-frequency waveform sampling sequence are identified and extracted, and mapped and correlated with the occurrence times of the undervoltage event and load change event recorded in the key data block. Finally, after time-domain alignment and semantic mapping processing, the high-frequency sampling sequence and the key data block are merged into a preliminary time-domain aligned dataset. The above scheme achieves cross-scale time alignment between high-frequency sampling sequences and frozen data from the energy meter through dual calibration of the clock source time stamp and the energy meter event timestamp, forming a unified time-domain reference system. The electrical event points in the sampling sequence and the recorded events in key data blocks are linked by timestamp mapping, establishing a correlation accurate to the sampling period. The high-frequency waveform sampling sequence provides a continuous waveform image composed of instantaneous voltage / current values, while the key data blocks provide discrete event records. The fusion of these two allows for: a quantitative correspondence between the waveform characteristics of electrical events, such as the magnitude and duration of sudden drops, and event types, such as voltage loss and load changes; the ability to trace the original waveforms in the load curve matrix data; verification of the timing logic of event records through the high-frequency sampling sequence; and the resolution of timing errors caused by clock drift in different subsystems through time stamp recalibration and consistency verification. This ensures that all data units have a unified absolute time reference, and the time correlation error between event records and waveform sampling is less than one sampling period. It also allows for the reconstruction of the complete power parameter state, instantaneous voltage / current values, and event markers at any given time. Ultimately, the spatiotemporal consistency fusion of multi-source heterogeneous data from electricity meters was achieved, providing underlying data support with complete temporal correlation for subsequent transient process analysis and event tracing and diagnosis.

[0029] Example 4: Figure 4 As shown, based on Example 1, the process of outputting specific wiring error types and power correction coefficients provided in this embodiment of the invention includes the following steps: S201: Using the high-frequency waveform sequence in the preliminary dataset as an independent real power source reference, calculate the theoretical value of the instantaneous power of the line; combine the given current transformer ratio and polarity parameters to calibrate and convert the current data in the high-frequency waveform; then compare and integrate the calibrated theoretical value of instantaneous power with the power value of the load curve read from the internal energy meter in the preliminary dataset point by point; generate a comprehensive error index characterizing the accuracy of the metering device; S202: By performing discrete spectrum transformation on the high-frequency waveform sequence, the amplitude, phase and time-varying pattern of each harmonic are extracted to form an electricity consumption characteristic spectrum; the current electricity consumption characteristic spectrum is deeply matched and the deviation is calculated with the reference spectrum under the historical normal electricity consumption mode to identify and generate a feature identifier indicating the type of abnormal electricity load. S203: The comprehensive error index, abnormal feature identification, waveform phase relationship, and undervoltage record information in the preliminary dataset are sent together into a pre-built topology inference engine. The topology inference engine performs multi-dimensional fitting and probability calculation on the actual error mode, harmonic characteristics, event records and 96 models, and finally outputs the most likely wiring error type and the power correction coefficient for power compensation calculation.

[0030] Among them, the 96 wiring models are based on the full state space model of a three-phase four-wire metering system, including: voltage loop combinations, 4 types of abnormalities: phase loss, U / V / W missing; phase sequence error, 6 permutations and combinations, neutral line loose connection, phase-to-phase short circuit; current loop combinations; 24 types of abnormalities, CT polarity reverse connection, 8 combinations; CT phase misconnection, 6 combinations; current loop open circuit, 9 current shunting methods for electricity theft; and 68 types of compound faults: including all possible combinations of voltage / current loop cross faults, each combination corresponding to a unique power vector characteristic; The system processes the high-frequency voltage and current waveform sequences acquired non-contactly, generating a high-sampling-rate theoretical power value sequence using an instantaneous power algorithm. This theoretical value sequence is then calibrated and converted using externally provided current transformer ratio and polarity parameters, ensuring its representation is aligned with the electricity meter readings within the same dimension and reference system. Subsequently, the theoretical power sequence, now unified in this dimension, is compared with the load curve power value sequence read from the electricity meter's internal data. Point-by-point difference calculations are performed on a unified time axis, and the differences over the entire time interval are integrated to obtain a comprehensive error index characterizing the long-term operational deviation of the entire metering system. Simultaneously, another processing step performs discrete spectrum analysis on the same high-frequency waveform, extracting the amplitude, phase, and dynamic modes of each harmonic to form a spectral fingerprint of the current user's electricity consumption characteristics. This system performs multi-dimensional, in-depth matching calculations with a benchmark spectrum model trained based on the user's historical normal electricity consumption data. By quantifying the deviations between the two in specific frequency band energy distribution, phase symmetry, and modal stability, a characteristic identifier pointing to a specific abnormal load type is ultimately identified and generated.

[0031] The working principle and beneficial effects of the above technical solution are as follows: First, this embodiment uses the high-frequency waveform sequence in the preliminary dataset as an independent real power source reference to calculate the theoretical instantaneous power value of the line; then, combined with the given current transformer ratio and polarity parameters, the current data in the high-frequency waveform is calibrated and converted; next, the calibrated theoretical instantaneous power value is compared point-by-point with the load curve power value read from the internal energy meter in the preliminary dataset and integrated; a comprehensive error index characterizing the accuracy of the metering device is generated; secondly, by performing discrete spectrum transformation on the high-frequency waveform sequence, the amplitude, phase, and time-dependent properties of each harmonic are extracted. The changing patterns constitute an electricity consumption characteristic spectrum. The current electricity consumption characteristic spectrum is deeply matched and the deviation is calculated with the reference spectrum under the historical normal electricity consumption mode to identify and generate a feature identifier indicating the type of abnormal electricity load. Finally, the comprehensive error index, abnormal feature identifier and waveform phase relationship, undervoltage record and other information in the preliminary dataset are sent to a pre-built topology inference engine. The topology inference engine performs multi-dimensional fitting and probability calculation on the actual error pattern, harmonic characteristics, event records and 96 models, and finally outputs the most likely wiring error type and the power correction coefficient for power compensation calculation. The above scheme employs a combined time-frequency-topology analysis: high-frequency waveforms provide transient characteristics, harmonic analysis expands the frequency domain diagnostic dimension, and topological reasoning integrates steady-state and transient features to form a closed-loop detection logic; model-driven fault decoupling: 96 models cover the entire anomaly space, decoupling composite faults through vector features to avoid misjudgments by traditional threshold methods; dynamic error correction: comprehensive error indicators and correction coefficients are directly related to the physical causes of measurement errors, rather than statistical approximations; it achieves fully automated closed-loop processing from data acquisition and feature extraction to fault diagnosis, significantly improving the accuracy and traceability of measurement anomaly detection.

[0032] Example 5: Based on Example 4, the topology inference engine provided in this embodiment of the invention performs multi-dimensional fitting and probability calculation of the actual error patterns, harmonic features, event records, and 96 models, including the following steps: S2031: The core of the topology reasoning engine is a pre-generated and stored knowledge base of wiring errors, containing 96 predefined models of possible wiring errors. Each model describes the physical wiring method and also corresponds to a set of theoretical multi-dimensional features, namely the expected error range when an error occurs, specific harmonic distortion modes, phase angle offset rules, and typical voltage and current event sequences. S2032: The topology inference engine performs multi-dimensional fitting calculations, matching the multi-dimensional diagnostic feature vectors containing comprehensive error indices, abnormal feature identifiers, waveform phase relationships in the preliminary dataset, and undervoltage records with the theoretical feature vectors of all 96 models in the wiring error knowledge base in a full-space, multi-attribute synchronous manner; generating a comprehensive matching score for each model to quantify the degree of agreement between actual measurement data and the theoretical expectations of the model; S2033: Based on the comprehensive matching score, the topology inference engine initiates probability calculation and decision-making; the matching degree of all models is sorted and probabilistically processed, and the posterior probability of each model becoming the actual cause of the failure is calculated; the model with the highest posterior probability is selected as the diagnostic conclusion.

[0033] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, the core of the topology inference engine is a pre-generated and stored wiring error knowledge base, which contains 96 predefined models of possible wiring errors. Each model describes the physical wiring method and corresponds to a set of theoretical multi-dimensional features, namely the expected error range when an error occurs, specific harmonic distortion modes, phase angle offset rules, and typical voltage and current event sequences. The topology inference engine performs multi-dimensional fitting calculations, matching the multi-dimensional diagnostic feature vectors containing comprehensive error indicators, abnormal feature identifiers, waveform phase relationships in the preliminary dataset, and undervoltage records with the theoretical feature vectors of all 96 models in the wiring error knowledge base in a full-space, multi-attribute synchronous manner. A comprehensive matching score is generated for each model to quantify the degree of agreement between the actual measurement data and the theoretical expectations of the model. Finally, based on the comprehensive matching score, the topology inference engine initiates probability calculation and decision-making. The matching degrees of all models are sorted and probabilistically processed to calculate the posterior probability of each model becoming the actual cause of the fault. The model with the highest posterior probability is selected as the diagnostic conclusion. The above scheme offers the following advantages: efficient search and decision-making: pre-stored models in the knowledge base eliminate the need for real-time solving of complex physical equations during the diagnostic process, making the computational complexity of matching manageable; improved robustness: multi-dimensional feature fusion reduces the risk of misjudgment caused by single feature anomalies and improves tolerance to noise and interference; interpretable output: probabilistic results not only provide the final diagnosis but also quantify the confidence level of different fault modes, supporting subsequent manual review or system self-optimization; it achieves end-to-end automated reasoning from raw data to fault type, and its core advantage lies in transforming physical rules into a computable feature matching problem and balancing accuracy and robustness through probabilistic decision-making.

[0034] Example 6: Based on Example 5, the present invention provides a process for generating a comprehensive matching score for each model, which includes the following steps: S20321: Project and compare the multidimensional diagnostic feature vector with the theoretical expected feature vector of a certain model in the wiring error knowledge base in a high-dimensional feature space composed of multiple dimensions such as error, harmonics, phase and events. S20322: Assign different weight coefficients to each type of feature attribute according to its diagnostic importance, and calculate the deviation between the measured value and the theoretical value in each dimension respectively; S20323: Generate a normalized comprehensive matching score for this specific model by inversely synthesizing all weighted deviations.

[0035] The working principle and beneficial effects of the above technical solution are as follows: First, this embodiment projects and compares the multi-dimensional diagnostic feature vector with the theoretical expected feature vector of a model in the wiring error knowledge base within a high-dimensional feature space composed of multiple dimensions including error, harmonics, phase, and events. Second, it assigns different weight coefficients to each type of feature attribute according to its diagnostic importance and calculates the deviation between the measured and theoretical values ​​in each dimension. Finally, it generates a normalized comprehensive matching score for the specific model by inversely synthesizing all weighted deviations. The above solution features differentiated processing: weight allocation ensures that key diagnostic dimensions have a stronger impact on the score, reducing noise interference from secondary features; deviation balancing: weighted inverse synthesis avoids excessive influence of single-dimensional deviation on the overall score, improving matching robustness; standardized output: normalization ensures that the matching scores of different models are on the same scale, supporting subsequent probability calculation and ranking; combined with spatial comparison of physical features and statistical optimization methods, the matching score can accurately reflect the model's fit and adapt to the discrimination needs of different fault modes.

[0036] Example 7: Based on Example 6, the process for calculating the deviation between the measured value and the theoretical value in each dimension provided by this embodiment of the invention includes the following steps: S203221: By learning and training from a massive number of historical correct and incorrect wiring cases, weight coefficients are generated, and an inherent diagnostic significance coefficient is pre-set for each type of feature attribute; S203222: Assign the inherent diagnostic significance coefficient of the currently being compared feature attribute as the basis for weighting; use the difference measurement method based on statistical confidence interval to calculate the absolute difference between the measured value and the theoretical expected value of the feature attribute, and normalize the difference value with the standard deviation of the feature within the historical normal fluctuation range to obtain an original deviation. S203223: Multiply the original deviation by the assigned weight coefficient to generate a weighted deviation score for the current attribute dimension.

[0037] The working principle and beneficial effects of the above technical solution are as follows: First, this embodiment generates weight coefficients by learning and training from a large number of historical correct and incorrect wiring cases, and pre-sets an inherent diagnostic significance coefficient for each type of feature attribute; second, it assigns the corresponding inherent diagnostic significance coefficient as the basis for weighting the feature attribute currently being compared; it uses a difference measurement method based on statistical confidence intervals to calculate the absolute difference between the measured value and the theoretical expected value of the feature attribute, and normalizes the difference value with the standard deviation of the feature within the historical normal fluctuation range to obtain an original deviation; finally, it multiplies the original deviation by the assigned weight coefficient to generate a weighted deviation score for the current attribute dimension. The above scheme employs data-driven weight optimization: historical statistics ensure that weight coefficients are adapted to the actual fault distribution, improving diagnostic generalization ability; robustness difference measurement: standard deviation normalization suppresses misjudgments caused by measurement noise and normal fluctuations, enhancing algorithm stability; interpretable weighting: the explicit combination of weights and normalized deviation makes the scoring process transparent and traceable; the coupling of statistical learning and normalized difference calculation not only ensures the objectivity of deviation calculation, but also strengthens key features through weights, providing a high signal-to-noise ratio input for comprehensive matching score.

[0038] Example 8: Based on Example 7, the process for calculating the absolute difference between the measured value and the theoretical expected value of a feature attribute provided in this embodiment of the invention includes the following steps: S2032221: Call a pre-generated confidence interval parameter library, and form it through statistical analysis by analyzing the monitoring data of various characteristic attributes under massive historical normal wiring conditions; for each type of characteristic attribute, define the historical normal fluctuation range, which is expressed as a numerical interval with the mean as the center and positive and negative several times the standard deviation as the boundary, and a standard deviation benchmark value used to quantify the normal fluctuation amplitude of the characteristic. S2032222: Calculate the absolute difference between the measured value and the theoretical expected value; convert the absolute difference to the corresponding standard deviation benchmark value in the confidence interval parameter library, and assess the significance of the absolute difference relative to the normal fluctuation range; S20322223: Generate an original deviation degree representing the strength of relative differences through normalization processing based on historical statistical information.

[0039] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first calls a pre-generated confidence interval parameter library, which is formed by analyzing the monitoring data of various characteristic attributes under massive historical normal wiring conditions through statistical analysis; for each type of characteristic attribute, a historical normal fluctuation range is defined, which is a numerical interval centered on the mean and bounded by a certain number of standard deviations, and a standard deviation benchmark value used to quantify the normal fluctuation amplitude of the characteristic; secondly, the absolute difference between the measured value and the theoretical expected value is calculated; the absolute difference is proportionally converted to the corresponding standard deviation benchmark value in the confidence interval parameter library to evaluate the significance of the absolute difference relative to the normal fluctuation range; finally, through normalization processing based on historical statistical information, an original deviation degree characterizing the relative difference intensity is generated. The above scheme features standardized difference measurement: the confidence interval parameter library and standard deviation ratio conversion together ensure that the difference calculation is not affected by feature units or inherent fluctuations; dynamic robustness: the benchmark value based on historical statistics adapts to different operating environments, avoiding misjudgments caused by fixed thresholds; anomaly significance quantification: the normalized deviation directly maps the statistical anomaly degree of the difference, providing a high signal-to-noise ratio input for subsequent weighting and comprehensive scoring; the coupling of statistical benchmark and dynamic normalization ensures the objectivity of difference calculation and improves the stability of the algorithm under complex working conditions through historical data adaptation.

[0040] Example 9: Based on Example 8, the process of proportionally converting the absolute difference to the corresponding standard deviation benchmark value in the confidence interval parameter library provided in this embodiment of the invention includes the following steps: S20322221: A dynamic benchmark index group for extracting target feature attributes from a pre-established confidence interval parameter library, containing three core elements: historical mean center point, normal fluctuation boundary value, and standard deviation amplitude; automatically matching the current detected feature with the feature identifier code in the parameter library; S20322222: The attribute measurement values ​​obtained by feature extraction from the measured data stream are compared with the expected values ​​output by the theoretical model by algebraic difference operation to generate unstandardized raw difference quantities; the standard deviation amplitude in the dynamic benchmark index group is used as the calibration reference, and the raw difference quantities are converted into calibration difference coefficients that can be compared across features by establishing the difference-benchmark ratio relationship. S20322223: Based on the calibration difference coefficient, a second correction is performed using the normal fluctuation boundary values ​​recorded in the confidence interval parameter library, combined with the characteristics; when the calibration difference coefficient exceeds the fluctuation boundary, the amplitude compression mechanism is activated; when it is within the boundary range, the linear preservation mechanism is activated to retain the original proportional relationship; finally, a normalized deviation index with environmental adaptability is output.

[0041] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first extracts a dynamic benchmark index group of target feature attributes from a pre-established confidence interval parameter library, which includes three core elements: the center point of the historical mean, the normal fluctuation boundary value, and the standard deviation amplitude; it automatically matches the current detected feature with the feature identification code in the parameter library; secondly, it performs algebraic difference calculation between the attribute measurement value obtained by feature extraction from the measured data stream and the expected value output by the theoretical model to generate an unstandardized raw difference quantity; it uses the standard deviation amplitude in the dynamic benchmark index group as a calibration reference, and converts the raw difference quantity into a calibration difference coefficient that can be compared across features by establishing a difference-benchmark ratio relationship; finally, based on the calibration difference coefficient, it performs a secondary correction by combining the normal fluctuation boundary value recorded by the feature in the confidence interval parameter library; when the calibration difference coefficient exceeds the fluctuation boundary, the amplitude compression mechanism is activated; when it is within the boundary range, the linear preservation mechanism is activated to retain the original ratio relationship; finally, it outputs a normalized deviation index with environmental adaptability. The above scheme provides a feature-customized evaluation standard for subsequent processing through dynamic benchmark matching, establishes a unified comparison benchmark through difference standardization transformation, enhances the robustness of the system through adaptive correction mechanism, and forms a complete link from data acquisition to standardized output through three-stage processing; as a whole, it realizes the quantifiable comparison and evaluation of multi-source heterogeneous feature data.

[0042] Example 10: As Figure 5 As shown, based on Example 1, the process for automatically generating or updating inspection work orders provided by this embodiment of the invention includes the following steps: S301: The cloud marketing system receives and parses the test report, extracting three core judgment elements: comprehensive error exceeding standard mark, abnormal load feature code and wiring error type identifier; establishes an error feature vector group, including three dimensions: error exceeding standard level, harmonic pollution index and wiring error severity, with each dimension quantified into a five-level evaluation scale; S302: Based on the error feature vector group, call the preset work order to generate a decision tree; the decision tree contains three levels of judgment nodes: the first level node verifies whether the comprehensive error exceeds the legal threshold, the second level node analyzes whether the abnormal load has the characteristic pattern of electricity theft, and the third level node evaluates whether the wiring error causes the metering deviation; each node outputs a binary judgment result, forming a combination of work order triggering conditions; S303: When any two or more judgment conditions are met, the work order generation engine is activated; the work order generation engine automatically associates the historical audit records in the user's electricity consumption file to obtain the correlation index between the current anomaly and the historical records; for the first anomaly, a new work order is generated; for repeated anomaly patterns, the work order upgrade process is triggered, and a new anomaly event number and handling priority identifier are added to the original work order. S304: Automatically match the database of inspectors with the corresponding handling qualifications based on the wiring error type identifier, and generate the optimal personnel allocation plan by combining the work order priority identifier and the current task load of the personnel; for special work orders involving harmonic pollution, synchronously link the power quality management expert database resources.

[0043] The working principle and beneficial effects of the above technical solution are as follows: This embodiment forms a complete closed loop from anomaly feature extraction, condition judgment, work order generation, and task allocation, ensuring the automation and standardization of the audit process; through historical correlation analysis and work order upgrade mechanisms, the system has self-learning capabilities and continuously optimizes audit strategies; by combining anomaly types with personnel capabilities, optimal matching is achieved, improving audit accuracy and execution efficiency; the entire process is based on rules and quantitative analysis, reducing subjective judgment and enhancing the objectivity and authority of audit results; and it ensures that the generation, updating, and allocation of audit work orders meet business needs while possessing high efficiency, accuracy, and scalability.

[0044] Example 11: As Figure 6 As shown, based on Examples 1-10, the energy meter data acquisition and management system based on voltage-based non-contact anti-theft detection provided by this invention includes: The dataset generation module is used to collect real-time voltage and current waveform data on the line non-contactly through an insulating rod using a high-voltage clamp meter; it communicates with the energy meter through an infrared or RS485 interface to read key data such as freezing events, load curves and undervoltage records inside the energy meter, forming a preliminary dataset containing real-time waveforms of the line and data recorded inside the energy meter. The topology diagnostic module is used to calculate the comprehensive error of the energy meter and current transformer based on the collected preliminary dataset and the turns ratio and polarity parameters of the current transformer, and to judge the accuracy; to perform harmonic analysis on the collected real-time waveform of the line and identify abnormal power load characteristics; and to perform topology diagnostics, matching the actual wiring with 96 models and outputting the specific wiring error type and power correction coefficient. The report upload module is used to detect terminals to upload preliminary datasets, wiring error types, and electricity correction coefficients to the cloud marketing system via the 4G network; the cloud marketing system receives and stores the report, forms user electricity consumption profiles, and automatically generates or updates inspection work orders.

[0045] The working principle and beneficial effects of the above technical solution are as follows: The dataset generation module of this embodiment is used to collect real-time voltage and current waveform data on the line non-contactly through an insulating rod using a high-voltage clamp meter; it communicates with the electricity meter through an infrared or RS485 interface to read key data such as freezing events, load curves, and undervoltage records inside the electricity meter, forming a preliminary dataset containing real-time waveforms of the line and data recorded inside the electricity meter; the topology diagnosis module is used to calculate the comprehensive error of the electricity meter and the current transformer based on the collected preliminary dataset, combined with the transformation ratio and polarity parameters of the current transformer, and to judge the accuracy; it performs harmonic analysis on the collected real-time waveforms of the line to identify abnormal power load characteristics; at the same time, it performs topology diagnosis, matching the actual wiring with 96 models, and outputting specific wiring error types and power correction coefficients; the report upload module is used to upload the preliminary dataset and the detection report of specific wiring error types and power correction coefficients to the cloud marketing system via far-infrared communication through a 4G network or USB flash drive; the cloud marketing system receives and stores the detection report, forming part of the user's electricity consumption file, and automatically generates or updates inspection work orders. The dataset generation module of the above scheme simultaneously acquires real-time waveforms of the line and internal data from the electricity meter, constructing a complete information set including primary-side operating status and secondary-side metering data, providing a data foundation for subsequent analysis. The topology diagnostic module achieves comprehensive error assessment of the electricity meter and transformer through joint calculation of current transformer parameters and waveform data; simultaneously, based on harmonic characteristics and a pre-built model library, it completes wiring error detection and generates power correction coefficients, improving metering accuracy. Through harmonic analysis and load characteristic matching, abnormal electricity consumption patterns are identified from the perspective of waveform distortion and spectrum distribution, enhancing anti-theft detection capabilities. The report upload module directly connects the on-site inspection results with the cloud-based marketing system, automating the process of archiving inspection data, updating user profiles, and generating inspection work orders, forming a complete data chain from on-site diagnosis to management decision-making.

[0046] In summary, this embodiment achieves a technical closed loop of performance evaluation of metering equipment, abnormal power consumption detection, topology error correction, and management process optimization through modular functional division and data flow integration.

[0047] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of equivalents of this invention, this invention also intends to include these modifications and variations.

Claims

1. A method for data acquisition and management of electricity meters based on voltage-based non-contact anti-theft detection, characterized in that, Includes the following steps: Based on the collected preliminary dataset, combined with the turns ratio and polarity parameters of the current transformer, the comprehensive error of the energy meter and the current transformer is calculated to determine the accuracy; harmonic analysis is performed on the collected real-time waveform of the line to identify abnormal power load characteristics. Simultaneously, topology diagnostics are performed, matching the actual wiring with 96 models to output specific wiring error types and power correction coefficients.

2. The method for data acquisition and management of electricity meters based on non-contact voltage anti-theft detection as described in claim 1, characterized in that, The process of outputting specific wiring error types and power correction coefficients includes the following steps: Using the high-frequency waveform sequence in the preliminary dataset as an independent real power source reference, the theoretical value of the instantaneous power of the line is calculated; combined with the given current transformer ratio and polarity parameters, the current data in the high-frequency waveform is calibrated and converted. Then, the calibrated instantaneous power theoretical value is compared and integrated point by point with the load curve power value read from the internal energy meter in the preliminary data set; Generate a comprehensive error index characterizing the accuracy of the measuring device; By performing discrete spectrum transformation on the high-frequency waveform sequence, the amplitude, phase and time-varying pattern of each harmonic are extracted to form an electricity consumption characteristic spectrum; the current electricity consumption characteristic spectrum is deeply matched and the deviation is calculated with the reference spectrum under the historical normal electricity consumption mode to identify and generate a feature identifier indicating the type of abnormal electricity load. The comprehensive error index, abnormal feature identifier, waveform phase relationship and undervoltage record information in the preliminary dataset are fed into a pre-built topology inference engine. The topology inference engine performs multi-dimensional fitting and probability calculation on the actual error mode, harmonic characteristics, event records and 96 models, and finally outputs the most likely wiring error type and the power correction coefficient for power compensation calculation.

3. The method for data acquisition and management of electricity meters based on voltage-based non-contact anti-theft detection as described in claim 2, characterized in that, The topology inference engine performs multi-dimensional fitting and probability calculations on actual error patterns, harmonic characteristics, event records, and 96 models, including the following steps: The core of the topology reasoning engine is a pre-generated and stored knowledge base of wiring errors, containing 96 predefined models of possible wiring errors. Each model describes the physical wiring method and also corresponds to a set of theoretical multi-dimensional features, namely the expected error range, harmonic distortion mode, phase angle offset law, and typical voltage and current event sequence when the error occurs. The topology inference engine performs multi-dimensional fitting calculations, matching the input multi-dimensional diagnostic feature vector with the theoretical feature vectors of all 96 models in the wiring error knowledge base in a synchronous manner across the entire space and multiple attributes; it generates a comprehensive matching score for each model, quantifying the degree of agreement between actual measurement data and the theoretical expectations of the model; Based on the comprehensive matching score, the topology inference engine initiates probability calculation and decision-making; it sorts and probabilizes the matching degree of all models, calculates the posterior probability of each model becoming the actual cause of the failure, and selects the model with the highest posterior probability as the diagnostic conclusion.

4. The method for data acquisition and management of electricity meters based on voltage-based non-contact anti-theft detection as described in claim 3, characterized in that, The process of generating a comprehensive matching score for each model includes the following steps: The multidimensional diagnostic feature vector and the theoretical expected feature vector of a certain model in the wiring error knowledge base are projected and compared in a high-dimensional feature space composed of multiple dimensions such as error, harmonics, phase and events. Different weight coefficients are assigned to each type of feature attribute according to its diagnostic importance, and the deviation between the measured value and the theoretical value in each dimension is calculated respectively. By inversely synthesizing all weighted deviations, a normalized comprehensive matching score for the model is generated.

5. The method for data acquisition and management of electricity meters based on voltage-based non-contact anti-theft detection as described in claim 4, characterized in that, The process of calculating the deviation between measured and theoretical values ​​in each dimension includes the following steps: By learning and training from a massive number of historical correct and incorrect wiring cases, weight coefficients are generated, and an inherent diagnostic significance coefficient is pre-set for each type of feature attribute; Assign the corresponding inherent diagnostic significance coefficient to the feature attribute currently being compared as the basis for weighting; use the difference measurement method based on statistical confidence interval to calculate the absolute difference between the measured value and the theoretical expected value of the feature attribute, and normalize the difference value with the standard deviation of the feature within the historical normal fluctuation range to obtain an original deviation; Multiply the original deviation by the assigned weight coefficient to generate a weighted deviation score for the current attribute dimension.

6. The method for data acquisition and management of electricity meters based on non-contact voltage anti-theft detection as described in claim 5, characterized in that, The process of calculating the absolute difference between the measured value and the theoretical expected value of a feature attribute includes the following steps: A pre-generated confidence interval parameter library is called, and the data is formed by statistical analysis of the monitoring data of various characteristic attributes under massive historical normal wiring conditions. For each type of characteristic attribute, a historical normal fluctuation range is defined, which is a numerical interval centered on the mean and bounded by a number of standard deviations plus or minus, as well as a standard deviation benchmark value used to quantify the normal fluctuation amplitude of the characteristic. Calculate the absolute difference between the measured value and the theoretical expected value; convert the absolute difference proportionally to the corresponding standard deviation benchmark value in the confidence interval parameter library, and assess the significance of the absolute difference relative to the normal fluctuation range; By normalizing based on historical statistical information, an original deviation value representing the strength of relative differences is generated.

7. The method for data acquisition and management of electricity meters based on non-contact voltage anti-theft detection as described in claim 6, characterized in that, The process of proportionally converting the absolute difference to the corresponding standard deviation benchmark value in the confidence interval parameter library includes the following steps: A dynamic benchmark index set for target feature attributes is extracted from a pre-established confidence interval parameter library, containing three core elements: the center point of the historical mean, the boundary value of normal fluctuation, and the magnitude of the standard deviation; the current detected feature is automatically matched with the feature identifier code in the parameter library; The attribute measurements obtained by feature extraction from the measured data stream are compared with the expected values ​​output by the theoretical model using algebraic difference calculation to generate unstandardized raw difference quantities. The standard deviation amplitude in the dynamic benchmark index group is used as a calibration reference. By establishing the difference-benchmark ratio relationship, the raw difference quantities are converted into calibration difference coefficients that can be compared across features. Based on the calibration difference coefficient, a second correction is performed using the normal fluctuation boundary values ​​recorded in the confidence interval parameter library, combined with the characteristics; when the calibration difference coefficient exceeds the fluctuation boundary, the amplitude compression mechanism is activated. When the value is within the boundary range, a linear preservation mechanism is activated to maintain the original proportional relationship; the final output is a normalized deviation index that is environmentally adaptable.

8. The method for data acquisition and management of electricity meters based on non-contact voltage anti-theft detection as described in claim 1, characterized in that, It also includes using a high-voltage clamp meter to collect real-time voltage and current waveform data on the line without contact via an insulating rod; communicating with the electricity meter through an interface to read key data such as freezing events, load curves, and undervoltage records inside the electricity meter, forming a preliminary dataset containing real-time waveforms of the line and data recorded inside the electricity meter.

9. The method for data acquisition and management of electricity meters based on non-contact voltage anti-theft detection as described in claim 1, characterized in that, It also includes the detection terminal uploading preliminary datasets, wiring error types, and power correction coefficients to the cloud marketing system via a 4G network; the cloud marketing system receives and stores the reports, forms user power consumption profiles, and automatically generates or updates inspection work orders.

10. A data acquisition and management system for electricity meters based on voltage-based non-contact anti-theft detection, characterized in that, Include: The dataset generation module is used to collect real-time voltage and current waveform data on the line non-contactly through an insulating rod using a high-voltage clamp meter; it communicates with the electricity meter through an interface to read key data such as freezing events, load curves and undervoltage records inside the electricity meter, forming a preliminary dataset containing real-time waveforms of the line and data recorded inside the electricity meter. The topology diagnostic module is used to calculate the combined error of the energy meter and the current transformer based on the collected preliminary dataset and the turns ratio and polarity parameters of the current transformer, and to determine the accuracy. Harmonic analysis is performed on the collected real-time waveforms of the lines to identify abnormal power load characteristics; at the same time, topology diagnosis is performed to match the actual wiring with 96 models and output specific wiring error types and power correction coefficients. The report upload module is used to detect terminals to upload preliminary datasets, wiring error types, and electricity correction coefficients to the cloud marketing system via the 4G network; the cloud marketing system receives and stores the report, forms user electricity consumption profiles, and automatically generates or updates inspection work orders.

Images