A method for anti-electricity stealing based on abnormal connection of user single-phase electric meter

Abstract

The application discloses a method for anti-electricity stealing based on abnormal connection of user single-phase electric meter, and relates to the technical field of electric power metering, which comprises the following steps: installing sampling units on the incoming line side and the outgoing line side of the user single-phase electric meter, synchronously collecting single-phase voltage, current, neutral line current and temperature and humidity signals of the meter box, and generating synchronous electrical data sequences through standardization and time stamp alignment. Single-phase voltage and current fundamental phase angles are calculated based on voltage and current instantaneous value sequences, and a phase angle difference map is constructed. The effective value of the neutral line current is calculated, and an abnormal flag sequence is generated in combination with a normal load reference range. The environmental temperature and humidity abnormal fluctuation period is extracted. The above-processed features are input into a connection mode recognition model to identify suspected electricity stealing connection mode categories, improve the accuracy of meter connection abnormality recognition, and reduce misjudgment. The method can effectively identify meter connection abnormalities and improve the accuracy of electricity stealing identification.

Description

Technical Field

[0001] This invention belongs to the field of power metering technology, specifically a method for preventing electricity theft based on abnormal wiring of a user's single-phase electricity meter. Background Technology

[0002] Existing anti-electricity theft technologies mostly rely on voltage and current signals collected at a single point inside the user's single-phase electricity meter. They detect electricity theft by monitoring abnormalities in single electrical parameters such as power and power factor, or by simply comparing the difference between incoming and outgoing currents. The identification of meter wiring abnormalities is limited to amplitude deviation and does not delve into the phase relationship and environmental correlation. Conventional methods have shortcomings: they only focus on the total phase difference or power factor, without calculating and visualizing the difference for the fundamental phase angle of a single phase, making it difficult to capture subtle phase distortions; they ignore neutral current monitoring, while wiring abnormalities are often accompanied by an abnormal increase in neutral current; and they do not include environmental parameters such as meter box temperature and humidity as auxiliary criteria, making it impossible to distinguish between environmental interference and actual electricity theft, resulting in a high rate of missed and false detections.

[0003] The shortcomings of existing technologies point to two key issues that need to be addressed: the lack of a means to construct a spectrum based on the phase angle difference between the fundamental phase of single-phase voltage and current makes it impossible to accurately characterize the distribution features of single-phase phase anomalies; and the absence of a dynamic comparison mechanism between the effective value of the neutral current and the reference range of the normal load to generate anomaly indicators, as well as the lack of integration of abnormal fluctuations in ambient temperature and humidity with the phase angle difference spectrum for comprehensive judgment, makes it difficult to accurately identify the wiring patterns used for electricity theft. Summary of the Invention

[0004] This invention aims to solve at least one of the technical problems existing in the prior art;

[0005] Therefore, this invention proposes an anti-electricity theft method based on abnormal metering wiring of a user's single-phase electricity meter, comprising:

[0006] By installing sampling units on the incoming and outgoing sides of the user's single-phase electricity meter, the single-phase voltage signal, single-phase current signal, neutral line current signal, and temperature and humidity signals inside the meter box of the target electricity meter are collected simultaneously to form the original electrical quantity dataset.

[0007] The original electrical quantity dataset is standardized and timestamped to generate a time-aligned synchronous electrical data sequence, which includes a voltage instantaneous value sequence, a current instantaneous value sequence, a neutral line current instantaneous value sequence, and an ambient temperature and humidity sequence.

[0008] The fundamental phase angle of the single-phase voltage is calculated based on the instantaneous voltage value sequence, the fundamental phase angle of the single-phase current is calculated based on the instantaneous current value sequence, the effective value of the neutral line current is calculated based on the instantaneous neutral line current value sequence, and abnormal fluctuation periods in the environmental temperature and humidity sequence are extracted.

[0009] Based on the difference between the fundamental phase angle of the single-phase voltage and the fundamental phase angle of the single-phase current, a phase angle difference spectrum is constructed. At the same time, based on the effective value of the neutral line current and the normal load reference range, a neutral line abnormal current flag sequence is generated.

[0010] By combining the phase angle difference spectrum, the neutral line abnormal current marker sequence, and the abnormal fluctuation period, the suspected electricity theft wiring pattern category is identified through the wiring pattern identification model.

[0011] Furthermore, by installing sampling units on the incoming and outgoing sides of the user's single-phase electricity meter, the single-phase voltage signal, single-phase current signal, neutral current signal, and temperature and humidity signals inside the meter box of the target electricity meter are simultaneously collected, including:

[0012] A high-impedance voltage sampling circuit is connected in parallel to the live wire input terminal and the neutral wire input terminal of the user's single-phase electricity meter to obtain the single-phase voltage signal;

[0013] A low-impedance current sampling circuit is connected in series at the live wire output terminal of the user's single-phase electricity meter to obtain the single-phase current signal;

[0014] A closed-loop current transformer is installed on the total neutral line path of the user's single-phase electricity meter to obtain the neutral line current signal;

[0015] A probe integrating temperature and humidity sensing elements is fixedly deployed inside the meter box of a user's single-phase electricity meter to periodically acquire temperature and humidity signals inside the meter box;

[0016] Configure a unified clock source to provide synchronous trigger pulses for the sampling actions of all sampling units, ensuring the synchronization of timestamps for the single-phase voltage signal, single-phase current signal, neutral line current signal, and temperature and humidity signal inside the meter box.

[0017] Further, calculating the fundamental phase angle of the single-phase voltage based on the instantaneous voltage value sequence and calculating the fundamental phase angle of the single-phase current based on the instantaneous current value sequence includes:

[0018] A windowed Fourier transform is applied to the instantaneous voltage value sequence of a single phase to extract the real and imaginary parts of the fundamental frequency component. The phase angle of the fundamental frequency component of the single phase voltage is then calculated using the arctangent function.

[0019] The same windowed Fourier transform is applied to the instantaneous current value sequence of a single phase to extract the real and imaginary parts of the fundamental frequency component, and the phase angle of the fundamental frequency component of the single phase current is calculated by the arctangent function.

[0020] A sliding time window is used to smooth and filter the continuously calculated fundamental phase angles of the single-phase voltage and the single-phase current to eliminate phase jumps caused by random noise.

[0021] Furthermore, based on the difference between the fundamental phase angle of the single-phase voltage and the fundamental phase angle of the single-phase current, a phase angle difference spectrum is constructed, including:

[0022] Within a preset observation period, the difference between the fundamental phase angle of a single-phase voltage and the fundamental phase angle of the corresponding single-phase current at the same moment is calculated at fixed time intervals to form a sequence of instantaneous phase angle differences for a single phase.

[0023] The instantaneous phase angle difference sequence of the single phase is arranged in chronological order to form a single trajectory curve with time as the horizontal axis and phase angle difference value as the vertical axis;

[0024] Morphological analysis is performed on the single trajectory curve to extract the stable plateau intervals, abrupt change points, and periodic oscillation modes. The analysis results of the stable plateau intervals, abrupt change points, and periodic oscillation modes are integrated into structured data to form the phase difference map.

[0025] Furthermore, based on the effective value of the neutral current and the normal load reference range, a neutral abnormal current flag sequence is generated, including:

[0026] Within the preset effective value calculation window, the root mean square operation is performed on the instantaneous value sequence of the neutral line current to obtain continuous effective value data points of the neutral line current.

[0027] Retrieve a normal load reference range that matches the target user's electricity consumption type. The normal load reference range defines the theoretical allowable range of the effective value of the neutral line current under different total load power.

[0028] Each of the neutral line current RMS data points is compared with the theoretically permissible range under the total load power at the corresponding time.

[0029] If the effective value data point of the neutral line current continues to exceed the theoretically allowed range for a period of time exceeding a preset threshold, then a neutral line abnormal current flag is generated in the corresponding time period, and all neutral line abnormal current flags are arranged in chronological order to form the neutral line abnormal current flag sequence.

[0030] Furthermore, by combining the phase angle difference spectrum, the neutral line abnormal current marker sequence, and the abnormal fluctuation period, the wiring pattern identification model is used to determine the suspected electricity theft wiring pattern category, including:

[0031] The suspected electricity theft wiring patterns include reverse connection of live wire, reverse connection of neutral wire, interchange of live and neutral wires, and bypass wiring.

[0032] The stable plateau interval values ​​and abrupt change point information in the phase angle difference spectrum are encoded into feature vector one;

[0033] The flag density and flag duration segments in the neutral line abnormal current flag sequence are encoded into feature vector two;

[0034] The degree of deviation between the abnormal fluctuation period and the normal environmental parameters is encoded as feature vector three. The normal environmental parameters refer to the normal range of temperature and humidity inside the target meter box under historical normal operating conditions.

[0035] The first feature vector, the second feature vector, and the third feature vector are input into a pre-trained wiring pattern recognition network, and the wiring pattern recognition network outputs a probability distribution of belonging to different wiring pattern categories.

[0036] Select wiring pattern categories whose probability values ​​exceed a set threshold as the suspected electricity theft wiring pattern categories for output.

[0037] Furthermore, it also includes:

[0038] For the suspected electricity theft wiring pattern category, the corresponding electrical quantity verification rule set is invoked to quantify and verify the synchronous electrical data sequence, and a wiring anomaly diagnosis report containing anomaly type code, anomaly start timestamp, and anomaly confidence score is generated;

[0039] Based on the anomaly type code and anomaly confidence score in the wiring anomaly diagnosis report, different levels of on-site verification instructions are triggered, and the wiring anomaly diagnosis report and the on-site verification instructions are packaged into an anti-electricity theft handling task package.

[0040] The step of calling the corresponding electrical quantity verification rule set to quantify and verify the synchronous electrical data sequence for the suspected electricity theft wiring pattern category includes:

[0041] The pre-set set of electrical quantity verification rules for the live wire reverse connection mode requires verifying whether the phase difference between the voltage and current of the abnormal phase is close to 180 degrees, and at the same time verifying whether the electrical quantities of the non-abnormal phase remain normal.

[0042] The pre-set set of electrical quantity verification rules for the neutral line reverse connection mode requires verifying whether the ratio of the effective value of the neutral line current to the single-phase current vector and the theoretical value is greater than the preset deviation threshold, and at the same time verifying whether the single-phase voltage to ground potential offset exceeds the normal allowable range.

[0043] The pre-set set of electrical quantity verification rules for the live and neutral wire interchange mode requires verification of whether the phase relationship between voltage and current of all phases is systematically reversed, and verification of whether the neutral point potential is shifted.

[0044] The pre-set set of electrical quantity verification rules for bypass connection mode requires verifying whether the ratio of the metered current to the expected current estimated based on the user's load characteristics is lower than the preset current deviation threshold, and at the same time verifying whether the difference between the neutral current and the vector sum of the single-phase current exceeds the preset unbalanced current judgment threshold.

[0045] Using the data in the synchronous electrical data sequence, calculate the condition satisfaction degree of the corresponding electrical quantity verification rule set rule one by one, and perform a logical AND operation.

[0046] Furthermore, a wiring anomaly diagnostic report is generated, including the anomaly type code, the anomaly start timestamp, and the anomaly confidence score, including:

[0047] The suspected electricity theft wiring pattern categories that satisfy all verification rules are mapped to standardized exception type codes;

[0048] The earliest time point is selected as the anomaly start timestamp from the time point when the neutral line abnormal current flag sequence or the phase angle difference spectrum first shows obvious abnormality;

[0049] An anomaly confidence score is obtained by combining the probability value output by the wiring mode identification network, the weighted average of the satisfaction of each rule condition in the electrical quantity verification rule set, and the normalized index of the duration and intensity of the abnormal signal.

[0050] The abnormality type code, the abnormality start timestamp, and the abnormality confidence score, along with key data snapshots used for calculation, are filled into the report template according to a preset format to generate the wiring abnormality diagnosis report.

[0051] The key data snapshot refers to a segment of time-series data containing original voltage, current, temperature, humidity, and other specific measured values, extracted near the point of occurrence of the abnormal event corresponding to the abnormal start timestamp.

[0052] Furthermore, based on the anomaly type code and anomaly confidence score in the wiring anomaly diagnostic report, different levels of on-site verification instructions are triggered, including:

[0053] Set a confidence score threshold 1 and a confidence score threshold 2, wherein the confidence score threshold 2 is higher than the confidence score threshold 1;

[0054] When the abnormal confidence score is lower than the confidence score threshold, a low-level on-site inspection instruction is triggered. The low-level on-site inspection instruction only includes the requirement to perform routine inspections on the appearance and wiring of the meter.

[0055] When the abnormal confidence score is between the confidence score threshold one and the confidence score threshold two, a medium-level on-site verification instruction is triggered. The medium-level on-site verification instruction requires, on the basis of routine inspection, the use of portable verification equipment to re-verify the key electrical quantities.

[0056] When the abnormal confidence score is higher than the confidence score threshold two, a high-level on-site verification instruction is triggered. The high-level on-site verification instruction requires an immediate power outage inspection and the collection and recording of evidence for suspected electricity theft circuits.

[0057] Furthermore, the wiring anomaly diagnosis report and the on-site verification instruction are packaged into an anti-electricity theft handling task package, including:

[0058] Create a data structure containing a unique task identifier, the target meter asset number, and user information as the task package header;

[0059] The complete content of the wiring anomaly diagnosis report is embedded as the main data part of the anti-electricity theft handling task package;

[0060] The generated on-site verification instruction, including the instruction level, a list of specific verification items, and safety operation instructions, will be embedded in the anti-electricity theft handling task package as part of the operation guidance.

[0061] The complete anti-electricity theft response task package data is digitally signed and timestamped to form the final anti-electricity theft response task package that can be distributed to the field mobile operation terminal.

[0062] Compared with the prior art, the beneficial effects of the present invention are:

[0063] By calculating the fundamental phase angle of single-phase voltage from the instantaneous voltage value sequence and the fundamental phase angle of single-phase current from the instantaneous current value sequence, a phase angle difference map is constructed based on the difference between the corresponding single phases. This map presents the degree of deviation between the fundamental phase angles of single-phase voltage and current in terms of time series or phase sequence. Compared with conventional methods that only calculate the total phase difference or power factor, it can finely characterize the independent distribution features of single-phase phase anomalies, making the subtle phase distortions of single-phase or multi-phase phases that were originally hidden in the total phase difference explicit. This improves the sensitivity of identifying phase shifts caused by metering wiring errors and enables accurate location of wiring anomalies.

[0064] The effective value is calculated from the instantaneous value sequence of the neutral current, and an abnormal neutral current flag sequence is generated by combining it with the normal load reference range. Simultaneously, abnormal fluctuation periods in the ambient temperature and humidity sequence are extracted. The phase angle difference spectrum, the abnormal neutral current flag sequence, and the abnormal temperature and humidity periods are fused and input into the wiring pattern identification model. Comparison of the effective value of the neutral current with the normal load reference range can capture abnormal increases in neutral current that are overlooked in conventional anti-theft measures, compensating for the blind spots of single-phase current monitoring. The introduction of abnormal temperature and humidity fluctuation periods helps to eliminate the influence of environmental interference on electrical parameters. Compared with single electrical parameter threshold judgment, multi-feature fusion discrimination comprehensively reflects the multi-dimensional performance of wiring anomalies in phase relationship, neutral current, and environmental correlation. This makes the model's classification of electricity theft wiring patterns more closely match actual anomaly characteristics, reduces misjudgments caused by false alarms due to single features, and improves the overall accuracy of the discrimination. Attached Figure Description

[0065] Figure 1 This is a flowchart illustrating the steps of an anti-electricity theft method based on abnormal metering wiring of a user's single-phase electricity meter, as described in this invention.

[0066] Figure 2 A flowchart for the synchronous acquisition of signals by the sampling unit;

[0067] Figure 3 A flowchart for constructing a phase difference map;

[0068] Figure 4 Phase angle difference spectrum;

[0069] Figure 5 This is a diagram showing the analysis of abnormal neutral line current. Detailed Implementation

[0070] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0071] See Figure 1By deploying sampling units on the incoming and outgoing sides of the user's single-phase electricity meter, the system synchronously collects single-phase voltage signals, single-phase current signals, neutral current signals, and temperature and humidity signals within the meter box. These raw signals constitute the raw electrical quantity dataset. This raw electrical quantity dataset undergoes standardization and timestamp alignment processing to generate a synchronous electrical data sequence that is strictly aligned in the time dimension. This sequence includes a voltage instantaneous value sequence, a current instantaneous value sequence, a neutral current instantaneous value sequence, and an ambient temperature and humidity sequence. The fundamental phase angle of the single-phase voltage is calculated based on the voltage instantaneous value sequence, the fundamental phase angle of the single-phase current is calculated based on the current instantaneous value sequence, and the effective value of the neutral current is calculated based on the neutral current instantaneous value sequence. Abnormal fluctuation periods are extracted from the ambient temperature and humidity sequence. A phase angle difference map is constructed based on the difference between the fundamental phase angles of the single-phase voltage and the single-phase current. Simultaneously, an abnormal neutral current marker sequence is generated based on the effective value of the neutral current and the normal load reference range. By combining the phase angle difference spectrum, the abnormal current marker sequence of the neutral line, and the abnormal fluctuation period, the suspected electricity theft wiring pattern category is identified through the wiring pattern identification model.

[0072] See Figure 2 In one embodiment of the invention, the signal acquisition unit needs to be installed according to a specific physical connection method. A high-impedance voltage sampling circuit is connected in parallel to the live wire and neutral wire input terminals of the user's single-phase electricity meter. The impedance design of this high-impedance voltage sampling circuit needs to be much larger than the impedance of the meter's internal circuit to avoid the shunting effect on the original voltage signal after parallel connection, thereby obtaining a single-phase voltage signal that can truly reflect the voltage status of the input side. Similarly, a low-impedance current sampling circuit is connected in series to the live wire output terminal of the user's single-phase electricity meter. The impedance of this low-impedance current sampling circuit needs to be sufficiently small to ensure that no significant additional voltage drop is generated in the circuit after series connection, thereby obtaining the single-phase current signal flowing through the meter's metering coil. A closed-loop current transformer is installed on the total neutral wire path of the user's single-phase electricity meter. This current transformer adopts a closed-loop magnetic core structure and can measure the total current flowing through the neutral wire without contact, thereby obtaining the neutral wire current signal. A probe integrating temperature and humidity sensors is fixedly deployed inside the meter box. This probe periodically acquires the ambient temperature and humidity signals inside the meter box at a set sampling frequency. Configuring a unified clock source is crucial to ensuring data synchronization. The clock source generates high-precision synchronization trigger pulses, which are distributed to the voltage sampling circuit, current sampling circuit, current transformer, and temperature and humidity probe. This drives all sampling units to perform sampling operations under the same clock beat, ensuring that the final acquired single-phase voltage signal, single-phase current signal, neutral line current signal, and temperature and humidity signals inside the meter box have completely synchronized timestamps. The originally acquired signals together constitute the original electrical quantity dataset.

[0073] In some embodiments, the raw electrical quantity dataset needs to be preprocessed to generate the sequence required for subsequent calculations. Standardization is performed on the raw electrical quantity dataset to convert signal values ​​collected by different sampling units, which have different physical units and dimensions, into the same numerical range. Simultaneously with standardization, all signals are aligned in the time dimension based on a synchronization timestamp assigned by a unified clock source, eliminating misaligned data points that may be caused by communication delays, ultimately generating a synchronous electrical data sequence with strict time alignment. The synchronous electrical data sequence includes a sequence of instantaneous voltage values ​​obtained from single-phase voltage signal processing, a sequence of instantaneous current values ​​obtained from single-phase current signal processing, a sequence of instantaneous neutral current values ​​obtained from neutral line current signal processing, and an ambient temperature and humidity sequence obtained from temperature and humidity signals within the meter box.

[0074] It is understandable that the calculation of the fundamental phase angle relies on the spectral analysis of the instantaneous value sequence. A windowed Fourier transform is applied to the instantaneous voltage value sequence of a single phase. The process of the windowed Fourier transform involves first multiplying the instantaneous voltage value sequence by a finite-length window function, such as a Hanning window, to reduce spectral leakage. Then, a fast Fourier transform is performed on the windowed sequence, and the complex representation corresponding to the fundamental frequency component (i.e., its real and imaginary parts) is extracted from the transform result. The phase angle corresponding to this complex number is calculated using the arctangent function. Specifically, the imaginary part is divided by the real part to obtain the ratio, and the arctangent operation is performed on this ratio to obtain the fundamental phase angle of the phase voltage at the corresponding moment. The calculation of the fundamental phase angle of a single-phase voltage is performed independently for each instantaneous voltage value sequence using this method. The calculation of the fundamental phase angle of a single-phase current uses the exact same algorithm. The same windowed Fourier transform is applied to the instantaneous current value sequence of a single phase to extract the real and imaginary parts of its fundamental frequency component, and then the fundamental phase angle of the single-phase current is calculated using the arctangent function.

[0075] In some embodiments, to improve the stability of phase angle data, post-processing of the original calculation results is required. A sliding time window is used to smooth and filter the fundamental phase angles of the continuously calculated single-phase voltage and current. The sliding time window takes a fixed number of historical and future phase angle data points forward and backward from the current calculation point, for example, a rectangular window or Gaussian window of length N. The data points within the window are weighted and averaged, and the average value is used as the smoothed phase angle output for the current point. The smoothing filtering algorithm effectively suppresses random phase angle jumps caused by sampling noise, grid harmonics, or numerical fluctuations during calculation, resulting in a clearer trend and characteristics in the subsequently constructed phase angle difference spectrum. Its calculation formula can be expressed as:

[0076] ;

[0077] in: Indicates at time The fundamental phase angle of the voltage or current after smoothing and filtering Indicates at time The original fundamental phase angle was obtained by windowed Fourier transform and arctangent calculation. It is a summation index. The half-width of the sliding window determines the number of data points involved in the smoothing process. It is the time interval for phase angle calculation. It corresponds to the index. The window function weighting coefficients are such that the sum of all weighting coefficients is 1. Optionally, the above smoothing filtering process can be applied independently to the fundamental phase angle sequences of single-phase voltage and single-phase current.

[0078] See Figure 3 In one embodiment of the present invention, the construction of the phase difference map begins with the calculation of the instantaneous phase difference. Within a preset observation period, such as a complete 24-hour electricity consumption cycle, the difference between the fundamental phase angle of the single-phase voltage and the fundamental phase angle of the corresponding single-phase current at the same time is calculated at fixed time intervals. This difference directly reflects the power factor angle characteristic of the load of that phase. By calculating point by point, instantaneous phase difference sequences for phases A, B, and C are formed. Arranging the calculated instantaneous phase difference sequences of the single phases in chronological order can form three independent trajectory curves with time as the horizontal axis and the phase difference value as the vertical axis, representing the changes in the phase relationship of the loads of phases A, B, and C over time, respectively. Morphological analysis is performed on these single trajectory curves. The goal of morphological analysis is to extract the characteristic patterns contained in the curves, including identifying stable plateau intervals where the values ​​remain relatively constant, locating the abrupt change points where the phase difference value changes abruptly, and identifying whether there are periodic oscillation modes with a period of power frequency or other frequencies. The extracted stable plateau interval values, mutation point locations, and periodic oscillation modes are integrated into structured data containing fields such as timestamps, feature types, feature parameters, and curve relationships. This structured data constitutes the phase difference map.

[0079] In some embodiments, the generation of the neutral line abnormal current flag sequence depends on the calculation and comparison of the effective value of the neutral line current. Within a preset effective value calculation window, for example, with a window length of 1 minute and a sliding step of 30 seconds, the root mean square operation is performed on the instantaneous value sequence of the neutral line current. The calculation process involves taking the square of all instantaneous sampled values ​​within each window, calculating their average value, and then taking the square root of the average value to obtain continuous neutral line current effective value data points. A normal load reference range matching the target user's electricity consumption type is retrieved from a preset database or configuration file. Electricity consumption types include different types such as residential electricity, commercial electricity, and industrial electricity. The normal load reference range defines the theoretically permissible range of the effective value of the neutral line current under different total load power, taking into account load characteristics. Each calculated neutral line current effective value data point is compared with the theoretically permissible range obtained by querying the normal load reference range based on the total load power at the corresponding time. If a neutral current RMS data point exceeds the theoretically allowed range for a period of time exceeding a preset time threshold, for example, if it exceeds the limit for 10 consecutive RMS calculation windows, a neutral current abnormality flag is generated in the corresponding time period. This flag includes the start time, end time, and degree of exceedance of the abnormality. All generated neutral current abnormality flags are arranged in chronological order to form a neutral current abnormality flag sequence.

[0080] It is understandable that constructing a phase difference map requires quantitative analysis of the curve morphology. Extracting stable plateau intervals from the curve can employ a variance-based judgment method. The variance of the phase difference data within a sliding window is calculated, and when the variance is continuously below a set threshold, the interval is determined to be a stable plateau interval. Locating abrupt change points can utilize differential or edge detection algorithms to find points where the phase difference value changes significantly between adjacent time points. Identifying periodic oscillation modes can be achieved by performing power spectrum analysis on data from suspected oscillation intervals to check for significant power frequency or its harmonic peaks. Morphological analysis of a single trajectory curve is performed to extract stable plateau intervals, abrupt change points, and periodic oscillation modes. These analytical results are then integrated into structured data; this integration process can be achieved by constructing a multi-dimensional feature vector.

[0081] In some embodiments, the definition of the normal load reference range has a clear calculation basis. The theoretical allowable upper and lower bounds of the normal load reference range are calculated based on the amplitude of the single-phase current combined with the normal measurement error range. The width of the theoretical allowable range can be functionally related to the total load power, and the allowable deviation band is appropriately widened as the load increases. Each neutral current RMS value data point is compared with the theoretical allowable range under the corresponding total load power. The comparison operation determines whether the data point value falls outside the closed interval formed by the lower and upper bounds of the theoretical allowable range. Optionally, the calculation of the neutral current RMS value can use an overlapping sliding window to improve the time resolution. The root mean square operation is performed on the instantaneous value sequence of the neutral current, and its specific calculation expression is:

[0082] ;

[0083] in: Indicates the first The effective value data points of the neutral line current output by the effective value calculation window. Indicates at time The instantaneous value of the neutral line current was collected. This represents the total number of instantaneous value sampling points contained within a single valid value calculation window. Indicates the reference time from which the calculation begins. This represents the step size of the sliding window, which is the time interval between the starting points of the calculation window for two adjacent valid values. This represents the sampling time interval of the sequence of instantaneous values ​​of the original neutral line current. For window indexing, This is the index of the sampling points within the window. Optional, if the sliding step size... If the time is less than the window length, the calculated effective value data points of the neutral line current overlap in time, which helps to detect the onset of anomalies more sensitively.

[0084] In one embodiment of the present invention, the suspected electricity theft wiring pattern category is predefined into several specific types, including live wire reverse connection, neutral wire reverse connection, live and neutral wire interchange, and bypass wiring. The process of encoding feature vector one involves extracting quantitative features from the structured data contained in the phase angle difference spectrum. The stable plateau interval values ​​in the phase angle difference spectrum describe the constant angle value maintained by the phase angle difference under normal or abnormal wiring conditions. The abrupt change information in the phase angle difference spectrum includes the time of the abrupt change, the difference in phase angle difference before and after the abrupt change, and the steepness of the abrupt change. These features, including the stable plateau interval values ​​and the abrupt change information, are arranged and quantified according to a predetermined order and dimension, and encoded as feature vector one. The process of encoding feature vector two targets the neutral line abnormal current marker sequence. The marker density in the neutral line abnormal current marker sequence refers to the number of abnormal markers appearing within a unit observation time, and the marker duration segment in the neutral line abnormal current marker sequence refers to the start and end times covered by each abnormal marker and the total duration. The marker density and marker duration segment are extracted from the marker sequence, and this information, such as density value, average duration, and longest duration, is encoded as feature vector two. The process of encoding feature vector three involves quantifying abnormal fluctuation periods. Abnormal fluctuation periods are periods in the environmental temperature and humidity sequence that deviate from the normal baseline or change regularly. The degree of deviation between abnormal fluctuation periods and normal environmental parameters is quantified. Normal environmental parameters can be the average value or range of temperature and humidity over a period of time. The degree of deviation can be calculated as the difference between the average temperature of the abnormal fluctuation period and the normal average temperature or the ratio of the fluctuation amplitude to the normal fluctuation amplitude. The degree of deviation is encoded as feature vector three. Normal environmental parameters refer to the normal range of temperature and humidity changes inside the target meter box under historical normal operating conditions.

[0085] In some embodiments, the input and output of the wiring pattern recognition network have explicit forms. Feature vector 1, feature vector 2, and feature vector 3 are input into a pre-trained wiring pattern recognition network, which is a machine learning model whose structure can be a deep neural network, support vector machine, or random forest, etc. The input layer dimension of the wiring pattern recognition network is equal to the total length of the concatenation of feature vector 1, feature vector 2, and feature vector 3. The wiring pattern recognition network undergoes several layers of nonlinear transformations or decision processes, ultimately producing a set of values ​​at the output layer. The output of the wiring pattern recognition network is a probability distribution of different wiring pattern categories. The number of neurons in the output layer is equal to the number of defined suspected electricity theft wiring pattern categories. Each neuron corresponds to one category, and the output value of each neuron represents the probability that the input feature combination belongs to that wiring pattern category. The sum of the probabilities of all categories is 1. Wiring pattern categories with probability values ​​exceeding a set threshold are selected as the final suspected electricity theft wiring pattern category output. The set threshold is a predefined value, such as 0.7. If the probability of one or more categories exceeds this threshold, all of these categories are output.

[0086] It is understandable that the training of the wiring pattern recognition network needs to be based on historical data. The pre-trained wiring pattern recognition network is obtained through supervised training on a large number of labeled samples. The labeled samples contain feature vector 1, feature vector 2, and feature vector 3 extracted from meter data with known wiring states, as well as the corresponding real wiring pattern category label. During training, the network continuously adjusts its internal parameters through backpropagation algorithm or other optimization methods to minimize the difference between the probability distribution of the network output and the real category label. The training objective of the wiring pattern recognition network is to establish a mapping relationship between feature vector 1, feature vector 2, and feature vector 3 and several wiring patterns, including reversed live wire connection, reversed neutral wire connection, live and neutral wire interchange, and bypass wiring.

[0087] In some embodiments, the encoding of feature vectors needs to consider the scale differences between features. When encoding the stable plateau interval values ​​and abrupt change information in the phase difference spectrum as feature vector one, different types of numerical features can be normalized, for example, the phase difference value can be normalized to between 0 and 1, and the abrupt change time point can be converted into a percentage of time relative to the start point of the observation period. When encoding the marker density and marker duration segment in the neutral line abnormal current marker sequence as feature vector two, the marker density can be calculated as the number of abnormal markers per hour, and the marker duration segment can be expressed as the duration length in minutes. When encoding the degree of deviation between the abnormal fluctuation period and the normal environmental parameters as feature vector three, the benchmark of the normal environmental parameters can be obtained by calculating the temperature and humidity statistics under the same period or similar operating conditions in historical data, and the degree of deviation can be quantified as a percentage of relative deviation.

[0088] Optionally, the wiring pattern recognition network can employ a softmax output layer to achieve a probability distribution output. Assuming the wiring pattern recognition network is a feedforward neural network with a single hidden layer, the probability distribution of its final output can be calculated using a softmax function of the following form:

[0089] ;

[0090] in: Represents the complete feature vector after concatenation. Under the condition that it is composed of eigenvector 1, eigenvector 2, and eigenvector 3, it belongs to the first... Each wiring mode category The probability of. It is an eigenvector The hidden layer feature vector is obtained after transformation by the nonlinear activation function of the hidden layer. It corresponds to the category The output layer weight vector, It corresponds to the category The output layer bias term. This refers to the total number of wiring mode categories, as shown in the embodiment. These correspond to reverse connection of live wire, reverse connection of neutral wire, interchange of live and neutral wires, and bypass connection, respectively. Representing vectors The transpose of . This represents the natural exponential function. It selects wiring pattern categories whose probability values ​​exceed a set threshold; specifically, it iterates through all... Output probability of each category Each probability value is then compared to a set threshold.

[0091] In one embodiment of the present invention, for the suspected electricity theft wiring pattern category output by the wiring pattern identification model, it is necessary to call the electrical quantity verification rule set pre-configured for each category to quantify and verify the synchronous electrical data sequence. The electrical quantity verification rule set pre-configured for the live wire reverse connection mode includes multiple verification rules. One verification rule requires verifying whether the phase difference between the voltage and current of the abnormal phase is close to 180 degrees. The criterion for being close to 180 degrees is that the absolute difference between the calculated phase difference value and 180 degrees is less than a preset phase tolerance threshold, such as 5 degrees. At the same time, another verification rule requires verifying whether the electrical quantities of the non-abnormal phase remain normal. The criterion for the non-abnormal phase electrical quantities remaining normal is that the phase difference between the voltage and current of the non-abnormal phase is within the normal power factor angle range of this type of load. The pre-set electrical quantity verification rule set for the neutral wire reverse connection mode also includes multiple verification rules. One of these rules requires verifying whether the proportion by which the effective value of the neutral current exceeds the effective value of the single-phase current is greater than a preset deviation threshold. This proportion is calculated by first calculating the effective value of the single-phase current based on the instantaneous value sequence of the single-phase current in the synchronous electrical data sequence, then subtracting this theoretical value from the measured effective value of the neutral current, and dividing the difference by the theoretical value to obtain the excess proportion. Another verification rule requires verifying whether the single-phase voltage-to-ground potential offset exceeds the normal allowable range. The voltage-to-ground potential offset can be obtained through measurement or calculation, and the normal allowable range is determined according to power supply system standards. The pre-set electrical quantity verification rule set for the live-neutral wire interchange mode requires verifying whether the phase relationship between the voltage and current of all phases is systematically reversed. The criterion for systematic reversal is that the phase relationship, which should originally be inductive or capacitive, is completely reversed after the live-neutral wire interchange. For example, the current phase of all phases changes from lagging behind the voltage to leading the voltage. It also verifies whether the neutral point potential has shifted, which can be determined by measuring the degree of voltage asymmetry. The pre-defined electrical quantity verification rule set for the bypass connection mode requires that the ratio of the metered current to the expected current estimated based on the user's load characteristics be lower than a preset current deviation threshold. User load characteristics can be estimated based on historical electricity consumption data or a model of similar users. Simultaneously, the difference between the neutral current and the vector sum of single-phase currents must exceed a preset unbalanced current judgment threshold. Using data from the synchronous electrical data sequence, including instantaneous voltage, current, and neutral current sequences, the condition satisfaction degree of each rule in the corresponding electrical quantity verification rule set is calculated. The condition satisfaction degree of each rule can be quantified as a value between 0 and 1, where 1 indicates complete satisfaction and 0 indicates complete non-satisfaction. A logical AND operation is performed on the condition satisfaction degrees of all rules within the same electrical quantity verification rule set. Numerically, this logical AND operation represents taking the minimum value of all condition satisfaction degrees. When the minimum value is greater than a judgment threshold, the rule set is considered to have passed the overall verification.

[0092] In some embodiments, generating a wiring anomaly diagnostic report requires converting the quantitative verification results into structured report information. Suspected wire theft wiring patterns that satisfy all verification rules are mapped to standardized anomaly type codes. This mapping process is completed using a predefined lookup table; for example, a live wire reverse connection pattern is mapped to code "F_RL," and a neutral wire reverse connection pattern is mapped to code "N_RL." Determining the anomaly start timestamp requires backtracking analysis. The earliest time point selected from the moments when a significant anomaly first appears in the neutral line abnormal current flag sequence or phase angle difference spectrum is taken as the anomaly start timestamp. For example, the start time of the first flag in the neutral line abnormal current flag sequence, or the time when a sudden change point is first detected in the phase angle difference spectrum, whichever is earlier. Calculating the anomaly confidence score requires integrating information from multiple dimensions, including the probability value output by the wiring pattern identification network, the weighted average of the satisfaction of each rule condition in the electrical quantity verification rule set, and the normalized index of the duration and intensity of the anomaly signal. The weighted average is the arithmetic mean of the satisfaction of each rule condition within the electrical quantity verification rule set after assigning different weights. The normalized index for the duration and intensity of abnormal signals can be quantified based on the proportion of the cumulative duration of the abnormal neutral current indicator to the observation period, and the average magnitude of the phase angle difference deviating from the normal baseline. The final value of the anomaly confidence score is calculated using a linear or nonlinear fusion function, with a value range between 0 and 100. The anomaly type code, anomaly start timestamp, and anomaly confidence score, along with a key data snapshot used for calculation (a time-series data segment containing original voltage, current, temperature, and humidity measurements taken near the anomaly start timestamp), are filled into a report template according to a preset format to generate a wiring anomaly diagnostic report. The report template defines the location, name, and data type of each field.

[0093] It is understandable that the rules in the electrical quantity verification rule set have explicit mathematical expressions or logical judgment processes. The calculation process for verifying that the ratio of the metered current to the expected current estimated based on the user's load characteristics is lower than a preset current deviation threshold can be expressed as: extracting the effective value of the metered current from the synchronous electrical data sequence. The expected effective current value can be calculated based on the user load characteristic model and the current voltage, or obtained by querying the load curve database. Calculate the ratio When the ratio Less than the preset current deviation threshold When the condition is met, the condition satisfaction level of the rule is 1; otherwise, the condition satisfaction level is 0 or decreases linearly according to the degree of deviation. The check verifies that the difference between the neutral current and the vector sum of the single-phase currents exceeds the preset unbalanced current judgment threshold. The calculation process requires first calculating the theoretical value of the vector sum of the single-phase currents. The calculation of this theoretical value requires vector synthesis of the instantaneous values ​​of the single-phase current, and then calculation of the absolute value of the difference. ,in This is the measured effective value of the neutral current. When the difference... The unbalanced current exceeds the preset threshold. When the condition is met, the rule's condition satisfaction is 1; otherwise, it is 0 or proportionally reduced.

[0094] In some embodiments, the anomaly confidence score can be calculated using a weighted fusion formula. Anomaly Confidence Score The calculation formula is as follows:

[0095] ;

[0096] in: This represents the final calculated anomaly confidence score. This represents the probability value output by the wiring pattern identification network, corresponding to the final identified category of the suspected electricity theft wiring pattern. This indicates the degree to which the conditions of each rule in the electrical quantity verification rule set are met. The weighted average, i.e. , It is the number of rules. It is the first The weight of each rule, . It is about the duration of abnormal signals With abnormal intensity The normalization function has an output value between 0 and 1. , , There are three weighting coefficients, and This is used to balance the contributions of different information sources to the confidence score. Key data snapshots used for calculation are typically embedded in the report as time-series data tables, see Table 1.

[0097] Table 1: Example Table of Key Data Snapshots During Abnormal Periods

[0098]

[0099] Optionally, the weighting coefficients for the anomaly confidence score can be adjusted based on the characteristics of different anomaly types. For reversed-wire modes, since the phase angle characteristics may be very significant, a weighting coefficient can be assigned... (Model probability weights) Higher values. For bypass wiring modes, the calculation results of the electrical quantity verification rule set may be more decisive, and can be assigned... (Rule satisfaction weight) Higher values. Duration of abnormal signals. With abnormal intensity normalization function Piecewise linear or logarithmic functions can be used to reflect the nonlinear effects of duration and intensity on confidence.

[0100] See Figure 4 In anti-theft methods based on abnormal wiring of single-phase electricity meters, the construction and analysis of phase angle difference maps are the core steps in identifying live wire reverse connection modes. Specifically, based on synchronous electrical data sequences, the fundamental phase angles of single-phase voltage and current are extracted using windowed Fourier transform. The instantaneous difference between the two is calculated and arranged in a time sequence, forming a trajectory curve with time as the horizontal axis and phase angle difference as the vertical axis. Morphological analysis of this curve extracts key features such as stable plateau intervals and abrupt change points: under normal operating conditions, the phase angle difference stabilizes at approximately 4.7 rad; when an anomaly occurs, the curve experiences a step change at the abrupt change point, entering a stable abnormal plateau interval, where the phase angle difference remains stable at approximately 3.14 rad. This feature is completely consistent with the physical mechanism of the systematic reversal of the voltage-current phase relationship under live wire reverse connection modes. Combining the neutral line abnormal current indicator sequence with periods of abnormal fluctuations in ambient temperature and humidity, a comprehensive identification and confidence assessment of the wiring mode can be further completed, providing quantitative basis for subsequent graded on-site verification.

[0101] In one embodiment of the present invention, two confidence score thresholds for classifying response levels need to be set, namely confidence score threshold one and confidence score threshold two, with confidence score threshold two set higher than confidence score threshold one. When the abnormal confidence score in the wiring abnormality diagnosis report is lower than confidence score threshold one, the system triggers a low-level on-site verification instruction. The low-level on-site verification instruction only includes the requirement to perform routine checks on the appearance and wiring of the meter. Routine check requirements may include visually checking whether the meter seal is intact, observing whether the wiring terminals have obvious loosening or burning marks, and verifying whether the incoming and outgoing line markings are clear and correct, etc. When the anomaly confidence score in the wiring anomaly diagnosis report falls between confidence score threshold one and confidence score threshold two, the system triggers a medium-level on-site verification instruction. This instruction, in addition to routine checks, requires the use of portable verification equipment to verify key electrical quantities. Portable verification equipment includes clamp meters, phase volt-amperes meters, etc. The verification of key electrical quantities may include on-site measurements of the voltage, current, phase angle, and neutral current at the meter's input terminals, which are then compared with the data in the diagnosis report. When the anomaly confidence score in the wiring anomaly diagnosis report exceeds confidence score threshold two, the system triggers a high-level on-site verification instruction. This instruction requires immediate power outage inspection, which must be performed after following proper power outage procedures. Evidence collection and recording of suspected electricity theft circuits are also required. This includes photographing or recording abnormal wiring methods, drawing on-site wiring diagrams, and extracting potentially tampered components as physical evidence.

[0102] In some embodiments, creating an anti-electricity theft response task package requires defining its data structure. A data structure containing a unique task identifier, target meter asset number, and user information is created as the header of the anti-electricity theft response task package. The unique task identifier is a globally unique string or number sequence used to track the task in the system. The target meter asset number is the meter's unique code in the asset management system. User information includes user name, address, contact information, etc. The complete content of the wiring anomaly diagnosis report is embedded as the main data part of the anti-electricity theft response task package. The complete content of the wiring anomaly diagnosis report includes the anomaly type code, anomaly start timestamp, anomaly confidence score, and key data snapshots. The triggered on-site verification instructions, including the instruction level, a list of specific verification items, and safety operation instructions, are embedded as the operation guidance part of the anti-electricity theft response task package. The instruction level directly corresponds to low, medium, or high level. The list of specific verification items details all checks and operation steps to be performed under the instruction level. The safety operation instructions include safety regulations, risk warnings, and emergency measures that must be followed during on-site operations. The complete anti-electricity theft response task package data is digitally signed. The digital signature is generated using an asymmetric encryption algorithm and a timestamp is added. The timestamp records the precise time when the anti-electricity theft response task package was packaged, forming a final anti-electricity theft response task package that can be distributed to the field mobile operation terminal. The field mobile operation terminal is equipped with a dedicated application that can receive, parse and display the contents of the anti-electricity theft response task package.

[0103] It is understandable that the setting of confidence score thresholds one and two needs to be based on business rules and historical handling experience. Confidence score threshold one can be set at a relatively low level, such as 30 points. Cases below this threshold are considered to have weak anomalous evidence and only require minimal visual inspection to rule out obvious misjudgments. Confidence score threshold two is set at a higher level, such as 70 points. Cases above this threshold are considered to have strong anomalous evidence and require immediate and stringent on-site intervention. Cases with anomaly confidence scores between confidence score threshold one and confidence score threshold two, such as scores between 40 and 65 points, are considered to have some suspicion but the evidence is not yet completely conclusive. Therefore, a medium-level on-site verification instruction is triggered, and on-site testing and verification are conducted using portable equipment to make a final judgment. Triggering different levels of on-site verification instructions is an automated decision-making process, and the decision logic is based on a simple comparison between the anomaly confidence score and the preset thresholds.

[0104] In some embodiments, the encapsulation format of the anti-electricity theft response task package needs to consider the efficiency of data transmission and parsing. The task package header, main data section, and operation instructions section can be organized according to a specific data serialization format, such as JSON or XML. The task unique identifier can be generated using a UUID algorithm. The target meter asset number and user information are obtained from the electricity marketing system or asset database. The content of the safety operation instructions is a fixed text template, with different templates selected according to different instruction levels. Digitally signing the complete task package data ensures that the task package is not tampered with during transmission and verifies the legitimacy of the task source. The digital signature is generated by encrypting the hash value of the task package data using the sender's private key. The attached timestamp is usually in Coordinated Universal Time (UTC) format. The final anti-electricity theft response task package is pushed to the designated field mobile operation terminal via a network communication module.

[0105] Optionally, the specific checklist for on-site inspection instructions can be dynamically refined based on the anomaly type code. For different anomaly type codes, such as the live wire reverse connection code "F_RL" or the bypass wiring code "BYPASS", the specific checklist for medium- and high-level on-site inspection instructions can include targeted check items in addition to the general check items. For live wire reverse connection, targeted check items might include using a phase volt-ammeter to measure the voltage and current phase relationship of the suspected reversed phase. For bypass wiring, targeted check items might include checking for any concealed additional wiring near the meter's inlet and outlet terminals. This dynamic refinement is achieved through a preset mapping rule that associates the anomaly type code with a set of additional inspection steps. Low-level on-site inspection instructions only include requirements for routine inspections of the meter's appearance and wiring; the list of routine inspection requirements is usually fixed and does not change with the anomaly type.

[0106] Integrity verification of the task package data before digital signature can be achieved by calculating its hash value. Let the complete data content of the task package be... Using a cryptographic hash function Calculate its hash value The digital signature process uses the sender's private key. For hash value Encrypt and generate a signature. The final distributed task package data body Includes raw data ,sign and timestamp ,Right now The mobile work terminal received... Then, use the sender's public key. For signature Decryption At the same time, recalculate the received hash value By comparison and Consistency is used to verify data integrity and signature validity. (Timestamp) Used to check the freshness of tasks and prevent old tasks from being replayed.

[0107] See Figure 5 This is a neutral current anomaly analysis chart. This chart is a core visualization result of the neutral current anomaly identification stage in anti-theft electricity detection. It accurately presents the abrupt change in the effective value of the neutral current from normal to continuously abnormal. From 15:12:30 to 15:13:15, the effective value of the neutral current stabilized in the range of 1.5A-2.2A, never exceeding the upper limit of the normal range (2A) marked by the dashed line, which is consistent with the theoretical allowable range of neutral current for this user's electricity consumption type. The current exhibits small periodic fluctuations, reflecting the single-phase imbalance characteristics under normal power load, with no abnormal abrupt changes. At 15:13:15, the current jumps instantaneously from approximately 2A to over 11A, exceeding the normal upper limit for the first time, which is the core candidate point for the anomaly start time stamp. After the abrupt change, the current oscillates regularly in the range of 6A-11A. All data points far exceed the normal upper limit, and the duration exceeds the preset threshold, meeting the anomaly judgment condition of "continuously exceeding the theoretical allowable range". The dots in the line graph precisely mark the peak and trough anomalies of each cycle, intuitively presenting the intensity and periodicity of the anomalies.

[0108] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for preventing electricity theft based on abnormal wiring of a user's single-phase electricity meter, characterized in that, The method includes: By installing sampling units on the incoming and outgoing sides of the user's single-phase electricity meter, the single-phase voltage signal, single-phase current signal, neutral line current signal, and temperature and humidity signals inside the meter box of the target electricity meter are collected simultaneously to form the original electrical quantity dataset. The original electrical quantity dataset is standardized and timestamped to generate a time-aligned synchronous electrical data sequence, which includes a voltage instantaneous value sequence, a current instantaneous value sequence, a neutral line current instantaneous value sequence, and an ambient temperature and humidity sequence. The fundamental phase angle of the single-phase voltage is calculated based on the instantaneous voltage value sequence, the fundamental phase angle of the single-phase current is calculated based on the instantaneous current value sequence, the effective value of the neutral line current is calculated based on the instantaneous neutral line current value sequence, and abnormal fluctuation periods in the environmental temperature and humidity sequence are extracted. Based on the difference between the fundamental phase angle of the single-phase voltage and the fundamental phase angle of the single-phase current, a phase angle difference spectrum is constructed. At the same time, based on the effective value of the neutral line current and the normal load reference range, a neutral line abnormal current flag sequence is generated. By combining the phase angle difference spectrum, the neutral line abnormal current marker sequence, and the abnormal fluctuation period, the suspected electricity theft wiring pattern category is identified through the wiring pattern identification model; Also includes: For the suspected electricity theft wiring pattern category, the corresponding electrical quantity verification rule set is invoked to quantify and verify the synchronous electrical data sequence, and a wiring anomaly diagnosis report containing anomaly type code, anomaly start timestamp, and anomaly confidence score is generated; Based on the anomaly type code and anomaly confidence score in the wiring anomaly diagnosis report, different levels of on-site verification instructions are triggered, and the wiring anomaly diagnosis report and the on-site verification instructions are packaged into an anti-electricity theft handling task package. The step of calling the corresponding electrical quantity verification rule set to quantify and verify the synchronous electrical data sequence for the suspected electricity theft wiring pattern category includes: The pre-set set of electrical quantity verification rules for the live wire reverse connection mode requires verifying whether the phase difference between the voltage and current of the abnormal phase is close to 180 degrees, and at the same time verifying whether the electrical quantities of the non-abnormal phase remain normal. The pre-set set of electrical quantity verification rules for the neutral line reverse connection mode requires verifying whether the ratio of the effective value of the neutral line current to the single-phase current vector and the theoretical value is greater than the preset deviation threshold, and at the same time verifying whether the single-phase voltage to ground potential offset exceeds the normal allowable range. The pre-set set of electrical quantity verification rules for the live and neutral wire interchange mode requires verification of whether the phase relationship between voltage and current of all phases is systematically reversed, and verification of whether the neutral point potential is shifted. The pre-set set of electrical quantity verification rules for bypass connection mode requires verifying whether the ratio of the metered current to the expected current estimated based on the user's load characteristics is lower than the preset current deviation threshold, and at the same time verifying whether the difference between the neutral current and the vector sum of the single-phase current exceeds the preset unbalanced current judgment threshold. Using the data in the synchronous electrical data sequence, calculate the condition satisfaction degree of the corresponding electrical quantity verification rule set rule one by one, and perform a logical AND operation; The anomaly confidence score is obtained by using the probability value output by the integrated wiring pattern identification network, the weighted average of the satisfaction of each rule condition in the electrical quantity verification rule set, and the normalized index of the duration and intensity of the anomaly signal.

2. The method for preventing electricity theft based on abnormal metering wiring of a user's single-phase electricity meter according to claim 1, characterized in that, Sampling units installed on the inlet and outlet sides of the user's single-phase electricity meter synchronously collect single-phase voltage signals, single-phase current signals, neutral line current signals, and temperature and humidity signals inside the meter box, including: A high-impedance voltage sampling circuit is connected in parallel to the live wire input terminal and the neutral wire input terminal of the user's single-phase electricity meter to obtain the single-phase voltage signal; A low-impedance current sampling circuit is connected in series at the live wire output terminal of the user's single-phase electricity meter to obtain the single-phase current signal; A closed-loop current transformer is installed on the total neutral line path of the user's single-phase electricity meter to obtain the neutral line current signal; A probe integrating temperature and humidity sensing elements is fixedly deployed inside the meter box of a user's single-phase electricity meter to periodically acquire temperature and humidity signals inside the meter box; Configure a unified clock source to provide synchronous trigger pulses for the sampling actions of all sampling units, ensuring the synchronization of timestamps for the single-phase voltage signal, single-phase current signal, neutral line current signal, and temperature and humidity signal inside the meter box.

3. The method for preventing electricity theft based on abnormal metering wiring of a user's single-phase electricity meter according to claim 2, characterized in that, Calculating the fundamental phase angle of a single-phase voltage based on the instantaneous voltage value sequence, and calculating the fundamental phase angle of a single-phase current based on the instantaneous current value sequence, including: A windowed Fourier transform is applied to the instantaneous voltage value sequence of a single phase to extract the real and imaginary parts of the fundamental frequency component. The phase angle of the fundamental frequency component of the single phase voltage is then calculated using the arctangent function. The same windowed Fourier transform is applied to the instantaneous current value sequence of a single phase to extract the real and imaginary parts of the fundamental frequency component, and the phase angle of the fundamental frequency component of the single phase current is calculated by the arctangent function. A sliding time window is used to smooth and filter the continuously calculated fundamental phase angles of the single-phase voltage and the single-phase current to eliminate phase jumps caused by random noise.

4. The method for preventing electricity theft based on abnormal metering wiring of a user's single-phase electricity meter according to claim 3, characterized in that, Based on the difference between the fundamental phase angle of the single-phase voltage and the fundamental phase angle of the single-phase current, a phase angle difference spectrum is constructed, including: Within a preset observation period, the difference between the fundamental phase angle of a single-phase voltage and the fundamental phase angle of the corresponding single-phase current at the same moment is calculated at fixed time intervals to form a sequence of instantaneous phase angle differences for a single phase. The instantaneous phase angle difference sequence of the single phase is arranged in chronological order to form a single trajectory curve with time as the horizontal axis and phase angle difference value as the vertical axis; Morphological analysis is performed on the single trajectory curve to extract the stable plateau intervals, abrupt change points, and periodic oscillation modes. The analysis results of the stable plateau intervals, abrupt change points, and periodic oscillation modes are integrated into structured data to form the phase difference map.

5. The method for preventing electricity theft based on abnormal metering wiring of a user's single-phase electricity meter according to claim 4, characterized in that, Based on the effective value of the neutral current and the normal load reference range, a neutral abnormal current flag sequence is generated, including: Within the preset effective value calculation window, the root mean square operation is performed on the instantaneous value sequence of the neutral line current to obtain continuous effective value data points of the neutral line current. Retrieve a normal load reference range that matches the target user's electricity consumption type. The normal load reference range defines the theoretical allowable range of the effective value of the neutral line current under different total load power. Each of the neutral line current RMS data points is compared with the theoretically permissible range under the total load power at the corresponding time. If the effective value data point of the neutral line current continues to exceed the theoretically allowed range for a period of time exceeding a preset threshold, then a neutral line abnormal current flag is generated in the corresponding time period, and all neutral line abnormal current flags are arranged in chronological order to form the neutral line abnormal current flag sequence.

6. The method for preventing electricity theft based on abnormal metering wiring of a user's single-phase electricity meter according to claim 5, characterized in that, Based on the phase angle difference spectrum, the neutral line abnormal current marker sequence, and the abnormal fluctuation period, the wiring pattern identification model identifies the suspected electricity theft wiring pattern category, including: The suspected electricity theft wiring patterns include reverse connection of live wire, reverse connection of neutral wire, interchange of live and neutral wires, and bypass wiring. The stable plateau interval values ​​and abrupt change point information in the phase angle difference spectrum are encoded into feature vector one; The flag density and flag duration segments in the neutral line abnormal current flag sequence are encoded into feature vector two; The degree of deviation between the abnormal fluctuation period and the normal environmental parameters is encoded as feature vector three. The normal environmental parameters refer to the normal range of temperature and humidity inside the target meter box under historical normal operating conditions. The first feature vector, the second feature vector, and the third feature vector are input into a pre-trained wiring pattern recognition network, and the wiring pattern recognition network outputs a probability distribution of belonging to different wiring pattern categories. Select wiring pattern categories whose probability values ​​exceed a set threshold as the suspected electricity theft wiring pattern categories for output.

7. The method for preventing electricity theft based on abnormal metering wiring of a user's single-phase electricity meter according to claim 6, characterized in that, Generate a wiring anomaly diagnostic report that includes the anomaly type code, anomaly start timestamp, and anomaly confidence score, including: The suspected electricity theft wiring pattern categories that satisfy all verification rules are mapped to standardized exception type codes; The earliest time point is selected as the anomaly start timestamp from the time point when the neutral line abnormal current flag sequence or the phase angle difference spectrum first shows obvious abnormality; The abnormality type code, the abnormality start timestamp, and the abnormality confidence score, along with key data snapshots used for calculation, are filled into the report template according to a preset format to generate the wiring abnormality diagnosis report. The key data snapshot refers to a segment of time-series data containing original voltage, current, temperature, humidity, and other specific measured values, extracted near the point of occurrence of the abnormal event corresponding to the abnormal start timestamp.

8. The method for preventing electricity theft based on abnormal metering wiring of a user's single-phase electricity meter according to claim 7, characterized in that, Based on the anomaly type code and anomaly confidence score in the wiring anomaly diagnostic report, different levels of on-site verification instructions are triggered, including: Set a confidence score threshold 1 and a confidence score threshold 2, wherein the confidence score threshold 2 is higher than the confidence score threshold 1; When the abnormal confidence score is lower than the confidence score threshold, a low-level on-site inspection instruction is triggered. The low-level on-site inspection instruction only includes the requirement to perform routine inspections on the appearance and wiring of the meter. When the abnormal confidence score is between the confidence score threshold one and the confidence score threshold two, a medium-level on-site verification instruction is triggered. The medium-level on-site verification instruction requires, on the basis of routine inspection, the use of portable verification equipment to re-verify the key electrical quantities. When the abnormal confidence score is higher than the confidence score threshold two, a high-level on-site verification instruction is triggered. The high-level on-site verification instruction requires an immediate power outage inspection and the collection and recording of evidence for suspected electricity theft circuits.

9. A method for preventing electricity theft based on abnormal wiring of a user's single-phase electricity meter according to claim 8, characterized in that, The wiring anomaly diagnosis report and the on-site verification instructions are packaged into an anti-electricity theft handling task package, including: Create a data structure containing a unique task identifier, the target meter asset number, and user information as the task package header; The complete content of the wiring anomaly diagnosis report is embedded as the main data part of the anti-electricity theft handling task package; The generated on-site verification instruction, including the instruction level, a list of specific verification items, and safety operation instructions, will be embedded in the anti-electricity theft handling task package as part of the operation guidance. The complete anti-electricity theft response task package data is digitally signed and timestamped to form the final anti-electricity theft response task package that can be distributed to the field mobile operation terminal.

Images