A smart meter box anti-electricity theft method and system

Abstract

The invention relates to the field of meter box monitoring, and discloses an intelligent meter box anti-electricity theft method and system, which are used to solve the problem of false alarms caused by harmonic interference when performing anti-electricity theft warning. The method and system include collecting time domain signals of current and voltage in the intelligent meter box within a historical detection time period, converting the time domain signals into frequency domain signals, extracting harmonic information of each harmonic, establishing a normal harmonic feature model, and obtaining a target harmonic range; collecting time domain signals of current and voltage in the intelligent meter box in real time, extracting harmonic information, performing filtering processing according to the target harmonic range, using an abnormality detection algorithm to perform a first abnormality judgment, and if the first judgment is abnormal, performing a multi-parameter joint evaluation, performing a second abnormality judgment according to the multi-parameter joint evaluation result, and performing an electricity theft warning according to the second abnormality judgment result, effectively reducing false alarms and missed alarms, and significantly enhancing the adaptability of the system to complex electricity theft modes.

Description

Technical Field

[0001] The present invention relates to the field of electric meter box monitoring, and more specifically to an intelligent electric meter box anti-electricity theft method and system. Background Art

[0002] The meter box anti-electricity theft technology is an important tool for identifying and preventing illegal electricity theft. It monitors the user's electricity usage behavior in real time through smart meters and sensors installed in the meter box. The technology uses the collected data such as current, voltage and power factor to detect anomalies through a preset algorithm, and then determines whether there is electricity theft, thereby effectively protecting power resources, reducing power losses, and ensuring the safe operation of the power grid.

[0003] For example, the corrosion-resistant, anti-slipping, anti-electricity-stealing junction box and its manufacturing method disclosed in the invention patent announcement number CN106959383A include: a junction box, a cover and a wiring structure; the junction box and the cover are both made of four-layer composite materials, a first wiring slot is provided at the bottom of the junction box, and a second wiring slot matching the first wiring slot is provided on the cover; the wiring structure is a socket-type integrated junction block, and the wiring socket integrated therein faces the first wiring slot; the wiring socket includes a socket body and a spinning structure, wherein the socket body includes a guide cone, a threaded sleeve and a socket for inserting a wire metal core, one end of the spinning structure is a tightening end, the middle part is a tightening thread matching the threaded sleeve, and the other end is a tightening cone with a taper smaller than the guide cone and a pressing tooth arranged inside; a photosensitive signal element is provided on one side of the integrated junction block facing the cover, which includes a battery, a photoresistor and a signal transmitting device. The present invention has the advantages of corrosion resistance, anti-aging, anti-slipping of the terminal, anti-electricity theft, and anti-interference.

[0004] For example, the announcement number is: CN106950409A invention patent announcement of the insulation waterproof anti-electricity theft metering integrated wiring device and its manufacturing method, including: including a junction box, a terminal, a conductor, an electric meter, and a signal acquisition terminal, wherein: it is also equipped with a cover; the junction box and the cover are four-layer composite materials, from the outside to the inside respectively, a rubber layer, a metal shielding layer, a carbon film and a plastic layer; the conductor is specifically a protective conductor, and the protective conductor structure is a protective sleeve, a surface shielding metal mesh, a secondary shielding wire mesh, and a core wire in sequence; the protective conductor is bundled to form a bus, and a bus protective sleeve is provided on the outer surface of the bus; a signal acquisition generating device is also provided; a battery and an action relay are provided in the signal acquisition generating device, and the action relay works when an abnormality occurs at the line connection end. The wiring device of the present invention has good insulation, strong waterproofness, can be used outdoors, is anti-electricity theft, and has precise electricity metering.

[0005] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:

[0006] Most of the existing anti-electricity theft methods are aimed at detecting traditional electricity theft methods, such as illegal wiring, bypassing the meter, reverse power backflow, etc. However, in many cases, nonlinear load devices are connected to the line, such as inverters, LED lights, UPS power supplies, and electric vehicle charging equipment. Due to their working characteristics, they will generate strong harmonic interference during operation, causing current and voltage waveform distortion. This harmonic interference may overlap with the judgment conditions of abnormal fluctuation detection in the anti-electricity theft system, triggering the wrong early warning mechanism, resulting in false alarms and missed alarms. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for preventing electricity theft in a smart meter box to solve the problems existing in the above-mentioned background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for preventing electricity theft in a smart meter box comprises the following steps: step 1: collecting time domain signals of current and voltage in a smart meter box within a historical detection time period, applying fast Fourier transform to convert the time domain signals into frequency domain signals, and extracting harmonic information of each harmonic according to the frequency domain signals, wherein the harmonic information includes harmonic amplitude, harmonic phase and harmonic frequency; step 2: establishing a normal harmonic characteristic model according to the harmonic information of each harmonic, and obtaining a target harmonic range; step 3: collecting time domain signals of current and voltage in the smart meter box in real time, converting the time domain signals into frequency domain signals, and extracting harmonic information, and using a digital filter to filter the harmonics within the target harmonic range according to the target harmonic range, suppressing the harmonic components, and retaining the fundamental signal; step 4: performing a first abnormality judgment using an abnormality detection algorithm improved by a residual analysis method; step 5: if the first judgment is abnormal, performing a multi-parameter joint evaluation; step 6: performing a second abnormality judgment according to the result of the multi-parameter joint evaluation, and performing an electricity theft warning according to the result of the second abnormality judgment.

[0010] Preferably, the application of fast Fourier transform to convert the time domain signal into a frequency domain signal, and the steps of extracting the harmonic information of each harmonic according to the frequency domain signal are: collecting the current and voltage signals of the users in the smart meter during the historical detection time period to obtain the time domain signal; converting the time domain signal into a frequency domain signal by fast Fourier transform; and extracting the harmonic amplitude and harmonic phase according to the frequency domain signal.

[0011] Preferably, the steps of establishing a normal harmonic feature model and obtaining a target harmonic range are as follows: performing statistical analysis on the harmonic information of each harmonic obtained to obtain a total harmonic distortion, a harmonic amplitude mean, and a harmonic amplitude standard deviation; setting a normal operating range for each harmonic amplitude; filtering out normal harmonic data according to the normal operating range, and establishing a normal harmonic feature model; using a K-means clustering algorithm to classify harmonic features, and according to the final classification cluster, calculating the range of sample features of each cluster as the characteristic pattern of the cluster, i.e., the target harmonic range.

[0012] Preferably, the steps of using the K-means clustering algorithm to classify harmonic features are as follows: Step 2.1: Use harmonic features as clustering features, all harmonic features as data sets, and the harmonic features in the data sets as data points; Step 2.2: Use the elbow method to determine the optimal number of clusters K; Step 2.3: Randomly select K data points in the data set as initial cluster centers, for each data point, calculate its Euclidean distance to each initial cluster center, for each data point, traverse the K initial cluster centers, and assign it to the cluster cluster corresponding to the nearest initial cluster center; Step 2.4: After traversing all data points, obtain initial cluster clusters, for each initial cluster cluster, calculate the mean of the data points in it to obtain a new cluster center; Step 2.5: Repeat steps 2.3 and 2.4 until the cluster center no longer changes, and obtain the final cluster cluster and the final classification cluster.

[0013] Preferably, the step of using a digital filter to filter the harmonics within the target harmonic range is: using a digital filter to filter the time domain signal and retain the fundamental signal, and if multiple harmonic frequencies need to be filtered, multiple digital filters are connected in series.

[0014] Preferably, the first abnormality judgment step of the abnormality detection algorithm improved by the residual analysis method is: according to the difference between the time domain signal and the signal after filtering, the residual of the signal before and after filtering is calculated; the dynamic residual threshold is set, if the residual of the signal before and after filtering is less than the dynamic residual threshold, it is judged to be normal, and no electricity theft behavior has occurred at present, and the process returns to continue to collect the time domain signals of the current and voltage in the smart meter box in real time; if the residual of the signal before and after filtering is greater than or equal to the dynamic residual threshold, it is judged to be abnormal, and a second abnormality judgment is performed.

[0015] Preferably, the step of setting the dynamic residual threshold is: extracting the historical residual sequence of the window in a sliding window to obtain the historical residual mean and the historical residual standard deviation; setting the initial residual threshold according to the historical residual mean and the historical residual standard deviation; extracting the current residual sequence of the window in real time to obtain the real-time residual mean and the real-time residual standard deviation; obtaining the real-time residual threshold according to the real-time residual mean and the real-time residual standard deviation, and the specific acquisition method is as follows: ; In the formula, Expressed as the real-time residual threshold, Expressed as the real-time residual mean, Expressed as the real-time residual standard deviation, is the sensitivity factor.

[0016] Preferably, the multi-parameter joint evaluation step is: obtaining the power factor of the smart meter box in real time; obtaining the total harmonic distortion rate of the smart meter box in real time; obtaining the total power supply and the total power of the user in real time, and obtaining the power balance error according to the total power supply and the total power of the user. The specific acquisition method is as follows: ; In the formula, Expressed as the energy balance error, Expressed as the total power supply, It is expressed as the metered electric energy of each user, where M is the total number of users. The instantaneous power of the smart meter box is collected in real time, the maximum instantaneous power and the minimum instantaneous power are obtained, and the average power is calculated. The power load fluctuation coefficient and power difference are calculated. The specific acquisition method is as follows: ; ; In the formula, Expressed as the power load fluctuation coefficient, Expressed as instantaneous power, Expressed as average power, Expressed as the maximum instantaneous power, Expressed as the minimum instantaneous power, It is expressed as power difference; the power theft risk index is obtained according to the power factor, total harmonic distortion rate, power balance error and power load fluctuation coefficient. The specific acquisition method is as follows: ; In the formula, Expressed as the electricity theft risk index, Expressed as the power load fluctuation coefficient, Expressed as the energy balance error, It is expressed as power factor, THD is expressed as total harmonic distortion, Expressed as power variability.

[0017] Preferably, the steps of performing a second abnormality judgment based on the result of the multi-parameter joint evaluation and issuing an electricity theft warning based on the result of the second abnormality judgment are: comparing the electricity theft risk index with a preset threshold; if the electricity theft risk index is less than the preset threshold, it is determined that no electricity theft behavior has occurred at present, and the process returns to continue to collect the time domain signals of the current and voltage in the smart meter box in real time; if the electricity theft risk index is greater than or equal to the preset threshold, it is determined that electricity theft behavior has occurred at present, and an early warning prompt is issued.

[0018] Preferably, a smart meter box anti-electricity theft system comprises: a historical data acquisition module, used to collect harmonic information of each harmonic of the smart meter box within a historical detection time period; a harmonic characteristic model establishment module, used to establish a normal harmonic characteristic model according to the harmonic information of each harmonic, and obtain a target harmonic range; a real-time data acquisition module, used to collect harmonics in the smart meter box in real time, and filter the harmonic information according to the target harmonic range; a first abnormality judgment module, used to perform a first abnormality judgment using an abnormality detection algorithm improved by a residual analysis method; a second abnormality judgment module, used to perform a multi-parameter joint evaluation and a second abnormality judgment; an early warning module, used to issue an early warning reminder for the abnormal situation of the second judgment, and remind relevant staff to conduct electricity theft inspection in time.

[0019] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0020] 1. Collect the time domain signals of current and voltage in the smart meter box during the historical detection period, apply fast Fourier transform to convert the time domain signals into frequency domain signals, extract the harmonic information of each harmonic according to the frequency domain signals, collect the time domain signals of current and voltage during the historical detection period, and extract the harmonic information of each harmonic through fast Fourier transform, which provides an accurate data basis for the anti-electricity theft solution of the smart meter box. Doing so can identify the harmonic characteristics of the equipment in operation, thereby effectively improving the accuracy and robustness of abnormal detection, reducing misjudgments and missed judgments, and optimizing the efficiency and reliability of the entire anti-electricity theft system.

[0021] 2. According to the harmonic information of each harmonic, a normal harmonic feature model is established to obtain the target harmonic range, accurately distinguish the harmonic characteristics of normal equipment operation from the abnormal harmonic characteristics, reduce the false alarm rate and missed alarm rate, and improve the accuracy of abnormal detection. At the same time, clarifying the target harmonic range can optimize the digital filter design and suppress abnormal harmonics more efficiently.

[0022] 3. Use the anomaly detection algorithm improved by the residual analysis method to make the first anomaly judgment, quickly identify the abnormal features in the signal that deviate from the normal mode, and provide efficient preliminary screening capabilities for the smart meter anti-electricity theft solution. Doing so can reduce the system's dependence on complex subsequent analysis and reduce the computational burden. At the same time, through the sensitivity to residuals, it can capture tiny abnormal signals, ensure that potential electricity theft or equipment anomalies can be discovered in a timely manner, and provide a reliable data foundation for subsequent multi-level detection and classification.

[0023] 4. Make a second abnormal judgment based on the results of the multi-parameter joint evaluation, and issue an electricity theft warning based on the results of the second abnormal judgment. By comprehensively considering multiple characteristic parameters, this step can effectively filter out false alarms, accurately locate the nature and source of abnormal behavior, and provide a clear basis for judging electricity theft. Combined with the early warning mechanism, the system can achieve rapid response and provide users and power companies with real-time and reliable electricity theft risk warnings, thereby optimizing system efficiency and reducing economic losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flow chart of a method for preventing electricity theft in a smart meter box provided in an embodiment of the present application.

[0025] Figure 2 A structural diagram of an intelligent meter box anti-electricity theft system provided in an embodiment of the present application.

[0026] Figure 3 Schematic diagram of the functional relationship between the residual, the time domain signal and the filtered signal involved in the embodiments of the present application. DETAILED DESCRIPTION

[0027] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures recorded in the following embodiments are merely illustrative. The anti-electricity theft method and system for a smart meter box involved in the present invention are not limited to the various structures recorded in the following embodiments. All other implementations obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0028] The present invention provides a method for preventing electricity theft from a smart meter box. Figure 1 As shown, the following steps are included:

[0029] Step 1: Collect the current and voltage signals in the smart meter box during the historical detection period to obtain the time domain signal sequence, apply fast Fourier transform to convert the time domain signal into the frequency domain signal, and extract the harmonic information of each harmonic according to the frequency domain signal. The harmonic information includes harmonic amplitude, harmonic phase and harmonic frequency. The detection period is manually set and can be 1 hour or 2 hours.

[0030] Fast Fourier Transform (FFT) is an algorithm for efficiently calculating Discrete Fourier Transform (DFT), which is used to convert discrete time-domain signals into frequency-domain signals. The core of FFT is to reduce the amount of calculation through divide-and-conquer method, so that it can quickly decompose the signal into components of different frequencies and extract amplitude and phase information. FFT is widely used in signal processing, power systems, communications and audio analysis for tasks such as harmonic analysis, spectrum analysis and filtering.

[0031] Time domain signals refer to the manifestation of signals changing over time, usually represented by time as the horizontal axis and signal amplitude as the vertical axis. It directly describes the instantaneous value of the signal, such as the magnitude of voltage or current at each moment. In the time domain, the overall waveform of the signal and its characteristics of change over time can be observed, but it is difficult to directly analyze the frequency components contained in the signal, so it is necessary to convert it into a frequency domain signal through Fourier transform for further analysis.

[0032] Frequency domain signals refer to the distribution of signals at different frequencies, usually represented by frequency as the horizontal axis and amplitude as the vertical axis. They are obtained from time domain signals through Fourier transform, revealing which frequency components the signal is composed of, as well as the amplitude and phase of these frequencies. Frequency domain signals are convenient for analyzing the periodic characteristics, harmonic components and noise interference of signals.

[0033] In this embodiment, it should be specifically explained that the time domain signal is converted into a frequency domain signal by applying a fast Fourier transform, and the steps of extracting harmonic information of each harmonic according to the frequency domain signal are as follows:

[0034] Collect the current and voltage signals of users in the smart meter during the historical detection period to obtain time domain signals;

[0035] Convert the time domain signal into a frequency domain signal. The specific acquisition method is as follows:

[0036] ;

[0037] In the formula, Represented as a time domain signal The complex form signal after conversion to the frequency domain, that is, the frequency domain signal, It is a time domain signal, which indicates the value of the original signal at the nth sampling point. It is the input data for fast Fourier transform. n is the index of the time domain point, and k is the index of the frequency point, which indicates the kth discrete frequency point in the frequency domain. The frequency corresponding to k is , is the actual frequency, is the sampling frequency, N is the number of sampling points, that is, for the time domain signal The total number of sampling points when performing fast Fourier transform, is the Fourier basis function, which is used to project the time domain signal onto different frequency components. is an imaginary unit, , in Fourier transform, is used to distinguish the sine and cosine components of a signal;

[0038] Extract the harmonic amplitude. The specific acquisition method is as follows:

[0039] ;

[0040] In the formula, It is expressed as the amplitude of the frequency domain signal at the kth point, reflecting the energy of the signal at this frequency component. Represented as the real component of the frequency domain signal, it represents the projection of the signal on the cosine component. It is represented as the imaginary component of the frequency domain signal, which represents the projection of the signal on the sinusoidal component. k is the discrete frequency point of the fast Fourier transform calculation result, which represents the kth frequency component in the signal.

[0041] Extract the harmonic phase. The specific acquisition method is as follows:

[0042] ;

[0043] In the formula, It is expressed as the phase angle of the kth point of the frequency domain signal, which represents the phase relationship of the frequency component relative to the time reference.

[0044] Obtaining harmonic information through fast Fourier transform can quickly and efficiently extract the frequency, amplitude and phase characteristics of the signal, revealing the distribution of fundamental and harmonics. It can identify the source of nonlinear loads, power quality problems or abnormal signals, provide clear spectrum analysis for signal processing, and facilitate further harmonic suppression, filter design or fault diagnosis.

[0045] Step 2: According to the harmonic information of each harmonic, a normal harmonic characteristic model is established to obtain the target harmonic range;

[0046] In this embodiment, it should be specifically explained that the steps of establishing a normal harmonic characteristic model and obtaining a target harmonic range are as follows:

[0047] The harmonic information of each harmonic is statistically analyzed to obtain the total harmonic distortion rate, harmonic amplitude mean value and harmonic amplitude standard deviation. The specific method of obtaining the total harmonic distortion rate is as follows: , where is the total harmonic distortion rate, is the amplitude of each harmonic, where , is the fundamental amplitude, is the number of the highest harmonic; for each subharmonic amplitude, set the normal operating range, and the harmonics beyond the range are abnormal signals. The fundamental wave is the fundamental frequency component of the signal, usually the lowest frequency component of the signal, and also the main frequency of the signal. The fundamental wave amplitude is the amplitude of the sine wave at the fundamental wave frequency, indicating the main energy distribution of the signal;

[0048] Normal harmonic data is screened out according to the normal operating range, and a normal harmonic feature model is established. The rows represent samples and the columns represent features. The features include harmonic order, harmonic amplitude range, harmonic amplitude mean, and harmonic amplitude standard deviation.

[0049] The K-means clustering algorithm is used to classify the harmonic features. According to the final classification cluster, the range of sample features of each cluster is calculated as the characteristic pattern of the cluster, that is, the target harmonic range. The pattern of each cluster represents a normal harmonic characteristic, and the statistical description of each pattern is output for subsequent classification or anomaly detection.

[0050] K-means clustering algorithm is an unsupervised learning algorithm used to divide data into K clusters so that the data points in each cluster have high similarity and the differences between different clusters are large. The algorithm allocates data points to the nearest cluster center through iterative optimization, and continuously adjusts the position of the cluster center until the cluster center is stable or reaches the set termination condition.

[0051] In this embodiment, it should be specifically explained that the steps of classifying harmonic features using the K-means clustering algorithm are as follows:

[0052] Step 2.1: Use the harmonic features as clustering features, all harmonic features as data sets, and the harmonic features in the data sets as data points;

[0053] Step 2.2: Use the elbow method to determine the optimal number of clusters K. The elbow method is a method for determining the optimal number of clusters K in a clustering algorithm. By evaluating the total sum of squared errors under different numbers of clusters, we can find the inflection point where the SSE decreases significantly. As the number of clusters K increases, the SSE will gradually decrease, but when K reaches a reasonable value, the decrease in SSE will slow down significantly. The K corresponding to the inflection point is the optimal number of clusters;

[0054] Step 2.3: Randomly select K data points in the data set as the initial cluster centers. For each data point, calculate its Euclidean distance to each initial cluster center. For each data point, traverse the K initial cluster centers and assign it to the cluster corresponding to the nearest initial cluster center. Euclidean distance is a method used to measure the shortest distance between two points in geometric space, which represents the straight-line distance between points. It calculates the difference between two points in each coordinate dimension, squares these differences, sums them, and then takes the square root to get the result;

[0055] Step 2.4: After traversing all data points, the initial clusters are obtained. For each initial cluster, the mean of the data points in it is calculated to obtain a new cluster center.

[0056] Step 2.5: Repeat steps 2.3 and 2.4 until the cluster center no longer changes, and obtain the final cluster and the final classification cluster.

[0057] Step 3: Collect the time domain signals of current and voltage in the smart meter box in real time, convert the time domain signals into frequency domain signals, and extract harmonic information. According to the target harmonic range, use a digital filter to filter the harmonics within the target harmonic range, suppress the harmonic components, and retain the fundamental signal, such as filtering the 5th and 7th harmonics;

[0058] In this embodiment, it should be specifically explained that the steps of using a digital filter to filter the harmonics within the target harmonic range are as follows:

[0059] Use digital filters to filter time domain signals Perform filtering and retain the fundamental signal. The specific acquisition method is: ,in is the filtered signal, is the time domain signal, is the transfer function of the digital filter, and if multiple harmonic frequencies need to be filtered, multiple digital filters are connected in series.

[0060] A digital filter is a digital or analog filter that is used to suppress signals within a specific frequency range (called a stopband) while allowing signals within other frequency ranges to pass through. It is usually designed with zeros and poles at the target frequency, and by precisely adjusting the filter parameters, it can achieve significant attenuation of specific frequencies (such as harmonics or noise) while having little effect on the fundamental and other frequencies.

[0061] Step 4: Use the anomaly detection algorithm improved by residual analysis method to make the first anomaly judgment;

[0062] In this embodiment, it should be specifically explained that the first abnormality judgment step of the abnormality detection algorithm improved by the residual analysis method is as follows:

[0063] Calculate the residual of the signal before and after filtering, such as Figure 3 The specific acquisition method is as follows:

[0064] ;

[0065] In the formula, It is expressed as the residual of the signal before and after filtering, is the time domain signal, is the filtered signal;

[0066] Set a dynamic residual threshold. If the residual of the signal before and after filtering is less than the dynamic residual threshold, it is judged as normal and no electricity theft is currently occurring. Return to continue collecting the time domain signals of the current and voltage in the smart meter box in real time. If the residual of the signal before and after filtering is greater than or equal to the dynamic residual threshold, it is judged as abnormal and electricity theft may be occurring. Perform a second abnormality judgment.

[0067] Table 1 Examples of residual data of signals before and after filtering

[0068]

[0069] As shown in Table 1, in a specific embodiment, the time domain signal is displayed in different dimensions. , filtered signal and the residual The table provides specific numerical data for the numerical changes at each sample point. The residual can be used to identify abnormal components that do not conform to the normal signal pattern. The size of the residual can reflect whether the filter effect in certain frequency bands is up to standard. In the anti-electricity theft solution for smart meters, calculating the residual helps capture abnormal harmonics or power characteristic changes caused by electricity theft.

[0070] Residual analysis is an anomaly detection method based on the difference between model predictions and actual observations. It determines whether the system is abnormal by analyzing the residuals (i.e., the difference between the observed value and the model predicted value). This method uses statistical properties, such as the mean, standard deviation, and distribution of the residuals, to establish a normal range or threshold, and detect whether the residuals deviate significantly from the normal range.

[0071] In this embodiment, it should be specifically explained that the step of setting the dynamic residual threshold is:

[0072] In a sliding window, such as the most recent 100 sampling points, extract the historical residual sequence of the window to obtain the historical residual mean and historical residual standard deviation;

[0073] The initial residual threshold is set according to the historical residual mean and historical residual standard deviation. The specific method of obtaining it is as follows:

[0074] ;

[0075] In the formula, Denotes the initial residual threshold, Expressed as the historical residual mean, Expressed as the historical residual standard deviation, is the sensitivity factor. In this embodiment, the sensitivity factor is 3, which means it is based on a normal distribution range of 99.7% and can be adjusted as needed;

[0076] Extract the current residual sequence of the window in real time to obtain the real-time residual mean and real-time residual standard deviation;

[0077] The real-time residual threshold is obtained based on the real-time residual mean and the real-time residual standard deviation. The specific acquisition method is as follows:

[0078] ;

[0079] In the formula, Expressed as the real-time residual threshold, Expressed as the real-time residual mean, Expressed as the real-time residual standard deviation, is the sensitivity factor.

[0080] The dynamic threshold can be adjusted in real time according to the fluctuation range of the signal to adapt to the characteristics of signal changes under nonlinear loads or environmental interference. When the signal fluctuation increases, the threshold is dynamically increased to avoid misjudging normal fluctuations as abnormalities. When the signal fluctuation decreases, the threshold is dynamically lowered to enhance the system's sensitivity to abnormalities. The fixed threshold may be misjudged as an abnormality when the signal fluctuation is large, increasing the false alarm rate. For example, during periods of high load fluctuations, the fixed threshold may not be able to tolerate the natural changes in the signal. The dynamic threshold is increased in real time according to the fluctuation range, making the system more tolerant to normal fluctuations. It is suitable for nonlinear load devices (such as inverters, charging piles) or scenarios with strong background noise.

[0081] Step 5: If the first judgment is abnormal, a multi-parameter joint evaluation is performed;

[0082] In this embodiment, it should be specifically explained that the steps of performing multi-parameter joint evaluation are:

[0083] Obtain the power factor of the smart meter box in real time. A low power factor may indicate abnormal load or harmonic interference.

[0084] Obtain the total harmonic distortion rate of the smart meter box in real time. The total harmonic distortion rate reflects the degree of harmonic interference. Usually, the value of electrical equipment is low, while non-linear loads may cause abnormal increases.

[0085] The total power supply and the total power of the user are obtained in real time, and the power balance error is obtained based on the total power supply and the total power of the user. The specific acquisition method is as follows:

[0086] ;

[0087] In the formula, Expressed as the energy balance error, Expressed as the total power supply, It is represented by the metered electric energy of each user, and M is the total number of users;

[0088] Collect the instantaneous power of the smart meter box in real time, obtain the maximum instantaneous power and the minimum instantaneous power, and calculate the average power, the power load fluctuation coefficient and the power difference. The specific acquisition method is as follows:

[0089] ;

[0090] ;

[0091] In the formula, Expressed as the power load fluctuation coefficient, Expressed as instantaneous power, Expressed as average power, Expressed as the maximum instantaneous power, Expressed as the minimum instantaneous power, It is expressed as power variability;

[0092] The power theft risk index is obtained based on the power factor, total harmonic distortion rate, power balance error and power load fluctuation coefficient. The specific method of obtaining it is as follows:

[0093] ;

[0094] In the formula, Expressed as the electricity theft risk index, Expressed as the power load fluctuation coefficient, Expressed as the energy balance error, It is expressed as power factor, THD is expressed as total harmonic distortion, Expressed as power variability.

[0095] The power factor is a parameter used to measure the efficiency of electric energy utilization in the power system. It reflects the ratio between active power and apparent power. The power factor value is between 0 and 1. A power factor close to 1 indicates that the electric energy is effectively utilized and the equipment is operating normally. A power factor close to 0 indicates that the electric energy utilization efficiency is low and there may be a large amount of reactive power or harmonic interference.

[0096] Power variability is used to measure the deviation between current power consumption and historical average power. It is usually used to detect short-term power consumption anomalies. Power theft may manifest as sudden abnormal changes in power, such as a sudden increase or decrease in power load. Power variability that significantly deviates from historical patterns (such as a sharp increase in instantaneous power) may be a signal of power theft.

[0097] Step 6: Perform a second abnormality judgment based on the multi-parameter joint evaluation results, and issue an electricity theft warning based on the second abnormality judgment results.

[0098] In this embodiment, it should be specifically explained that the steps of performing a second abnormality judgment based on the multi-parameter joint evaluation result and performing a power theft warning based on the second abnormality judgment result are as follows:

[0099] The electricity theft risk index is compared with the preset threshold. If the electricity theft risk index is less than the preset threshold, it is determined that no electricity theft has occurred and the system returns to continue collecting the time domain signals of the current and voltage in the smart meter box in real time. If the electricity theft risk index is greater than or equal to the preset threshold, it is determined that electricity theft has occurred and an early warning is issued to remind relevant staff to check the lines in a timely manner.

[0100] Through a hierarchical detection mechanism, rapid screening, intelligent classification and comprehensive verification are combined to achieve efficient and accurate identification of electricity theft. The first layer of preliminary screening uses an improved anomaly detection algorithm to quickly locate abnormal behaviors, while effectively filtering normal electricity users to reduce subsequent calculation pressure; the second layer of multi-parameter joint judgment comprehensively utilizes historical data, power consumption patterns, time characteristics, regional risks and other parameter verification results to further confirm the level of electricity theft risk and ensure the accuracy and credibility of the judgment. This process can effectively reduce false alarms and missed alarms, significantly enhance the system's adaptability to complex electricity theft patterns, and provide a clear basis for decision-making, optimize user experience, and protect the interests of power companies.

[0101] In this embodiment, it is necessary to specifically explain that a smart meter box anti-electricity theft system, such as Figure 2 As shown, the system includes:

[0102] A historical data collection module is used to collect harmonic information of each harmonic of the smart meter box within a historical detection period;

[0103] A harmonic characteristic model building module is used to build a normal harmonic characteristic model according to the harmonic information of each harmonic to obtain a target harmonic range;

[0104] Real-time data acquisition module, used to collect harmonics in the smart meter box in real time and filter the harmonic information according to the target harmonic range;

[0105] The first anomaly judgment module is used to perform the first anomaly judgment using an anomaly detection algorithm improved by residual analysis method;

[0106] The second abnormality judgment module is used to perform multi-parameter joint evaluation and perform the second abnormality judgment;

[0107] The early warning module is used to issue early warning reminders for the second abnormal situation and remind relevant staff to conduct electricity theft inspection in time.

[0108] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0109] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for preventing electricity theft in a smart meter box, characterized in that: The following steps are involved: Step 1: Collect the time domain signals of current and voltage in the smart meter box during the historical detection period, apply fast Fourier transform to convert the time domain signals into frequency domain signals, and extract the harmonic information of each harmonic according to the frequency domain signals. The harmonic information includes harmonic amplitude, harmonic phase and harmonic frequency. Step 2: According to the harmonic information of each harmonic, a normal harmonic characteristic model is established to obtain the target harmonic range; Step 3: Collect the time domain signals of current and voltage in the smart meter box in real time, convert the time domain signals into frequency domain signals, and extract harmonic information. According to the target harmonic range, use a digital filter to filter the harmonics within the target harmonic range, suppress the harmonic components, and retain the fundamental signal; Step 4: Use the anomaly detection algorithm improved by residual analysis method to make the first anomaly judgment; The steps of establishing a normal harmonic characteristic model and obtaining a target harmonic range are as follows: Perform statistical analysis on the harmonic information of each harmonic to obtain the total harmonic distortion rate, the mean value of the harmonic amplitude and the standard deviation of the harmonic amplitude; set the normal operating range for each harmonic amplitude; Filter out normal harmonic data according to the normal operating range and establish a normal harmonic characteristic model; The K-means clustering algorithm is used to classify the harmonic features. According to the final classification clusters, the range of sample features of each cluster is calculated as the characteristic pattern of the cluster, that is, the target harmonic range; The first abnormality judgment step of the abnormality detection algorithm improved by the residual analysis method is as follows: According to the difference between the time domain signal and the signal after filtering, the residual of the signal before and after filtering is calculated; Set a dynamic residual threshold. If the residual of the signal before and after filtering is less than the dynamic residual threshold, it is judged to be normal and no electricity theft is currently occurring. Return to continue to collect the time domain signals of the current and voltage in the smart meter box in real time. If the residual of the signal before and after filtering is greater than or equal to the dynamic residual threshold, it is judged to be abnormal and a second abnormality judgment is performed. Step 5: If the first judgment is abnormal, a multi-parameter joint evaluation is performed; The steps of performing multi-parameter joint evaluation are: Get the power factor of smart meter box in real time; Obtain the total harmonic distortion rate of the smart meter box in real time; The total power supply and the total power of the user are obtained in real time, and the power balance error is obtained based on the total power supply and the total power of the user. The specific acquisition method is as follows: Where ΔE is the energy balance error, E 供 Expressed as the total power supply, E 用户 It is represented by the metered electric energy of each user, and M is the total number of users; Collect the instantaneous power of the smart meter box in real time, obtain the maximum instantaneous power and the minimum instantaneous power, and calculate the average power, the power load fluctuation coefficient and the power difference. The specific acquisition method is as follows: In the formula, λ represents the power load fluctuation coefficient, P(t) represents the instantaneous power, and P avg It is expressed as average power, max(P(t)) is expressed as maximum instantaneous power, min(P(t)) is expressed as minimum instantaneous power, P diff It is expressed as power variability; The power theft risk index is obtained based on the power factor, total harmonic distortion rate, power balance error and power load fluctuation coefficient. The specific method of obtaining it is as follows: Where QD is the risk index of electricity theft, λ is the power load fluctuation coefficient, ΔE is the power balance error, cosφ is the power factor, THD is the total harmonic distortion rate, P diff It is expressed as power variability; Step 6: Perform a second abnormality judgment based on the multi-parameter joint evaluation results, and issue an electricity theft warning based on the second abnormality judgment results.

2. According to claim 1, a smart meter box anti-electricity theft method is characterized by: The step of converting the time domain signal into the frequency domain signal by applying the fast Fourier transform and extracting the harmonic information of each harmonic according to the frequency domain signal is as follows: Collect the current and voltage signals of users in the smart meter during the historical detection period to obtain time domain signals; Convert the time domain signal into the frequency domain signal through fast Fourier transform; Extract harmonic amplitude and harmonic phase based on frequency domain signals.

3. The method for preventing electricity theft by using a smart meter box according to claim 1, characterized in that: The steps of using the K-means clustering algorithm to classify harmonic features are as follows: Step 2.1: Use harmonic features as clustering features, all harmonic features as data sets, and the harmonic features in the data sets as data points; Step 2.2: Use the elbow method to determine the optimal number of clusters K; Step 2.3: Randomly select K data points in the data set as the initial cluster centers. For each data point, calculate its Euclidean distance to each initial cluster center. For each data point, traverse the K initial cluster centers and assign it to the cluster corresponding to the nearest initial cluster center. Step 2.4: After traversing all data points, the initial clusters are obtained. For each initial cluster, the mean of the data points in it is calculated to obtain a new cluster center. Step 2.5: Repeat steps 2.3 and 2.4 until the cluster center no longer changes, and obtain the final cluster and the final classification cluster.

4. The method for preventing electricity theft in a smart meter box according to claim 1, characterized in that: The steps of using a digital filter to filter harmonics within the target harmonic range are as follows: A digital filter is used to filter the time domain signal and retain the fundamental signal. If multiple harmonic frequencies need to be filtered, multiple digital filters are connected in series.

5. The method for preventing electricity theft by using a smart electric meter box according to claim 1, characterized in that: The steps of setting the dynamic residual threshold are: In a sliding window, extract the historical residual sequence of the window to obtain the historical residual mean and historical residual standard deviation; Set the initial residual threshold based on the historical residual mean and historical residual standard deviation; Extract the current residual sequence of the window in real time to obtain the real-time residual mean and real-time residual standard deviation; The real-time residual threshold is obtained based on the real-time residual mean and the real-time residual standard deviation. The specific acquisition method is as follows: Td t =avg t +r×bz t , In the formula, Td t Expressed as the real-time residual threshold, avg t Expressed as the real-time residual mean, bz t It is expressed as the real-time residual standard deviation, and r is the sensitivity factor.

6. The method for preventing electricity theft in a smart meter box according to claim 1, characterized in that: The steps of performing a second abnormality judgment based on the multi-parameter joint evaluation result and performing an electricity theft warning based on the second abnormality judgment result are as follows: The electricity theft risk index is compared with the preset threshold. If the electricity theft risk index is less than the preset threshold, it is determined that no electricity theft behavior has occurred at present, and the system returns to continue to collect the time domain signals of the current and voltage in the smart meter box in real time; If the electricity theft risk index is greater than or equal to the preset threshold, it is determined that electricity theft is currently occurring and an early warning is issued.

7. An intelligent meter box anti-electricity theft system, characterized by: The system comprises: A historical data collection module is used to collect harmonic information of each harmonic of the smart meter box within a historical detection period; A harmonic characteristic model building module is used to build a normal harmonic characteristic model according to the harmonic information of each harmonic to obtain a target harmonic range; Real-time data acquisition module, used to collect harmonics in the smart meter box in real time and filter the harmonic information according to the target harmonic range; The first anomaly judgment module is used to perform the first anomaly judgment using an anomaly detection algorithm improved by residual analysis method; The steps of establishing a normal harmonic characteristic model and obtaining a target harmonic range are as follows: Perform statistical analysis on the harmonic information of each harmonic to obtain the total harmonic distortion rate, the mean value of the harmonic amplitude and the standard deviation of the harmonic amplitude; set the normal operating range for each harmonic amplitude; Filter out normal harmonic data according to the normal operating range and establish a normal harmonic characteristic model; The K-means clustering algorithm is used to classify the harmonic features. According to the final classification clusters, the range of sample features of each cluster is calculated as the characteristic pattern of the cluster, that is, the target harmonic range; The first abnormality judgment step of the abnormality detection algorithm improved by the residual analysis method is as follows: According to the difference between the time domain signal and the signal after filtering, the residual of the signal before and after filtering is calculated; Set a dynamic residual threshold. If the residual of the signal before and after filtering is less than the dynamic residual threshold, it is judged to be normal and no electricity theft is currently occurring. Return to continue to collect the time domain signals of the current and voltage in the smart meter box in real time. If the residual of the signal before and after filtering is greater than or equal to the dynamic residual threshold, it is judged to be abnormal and a second abnormality judgment is performed. The second abnormality judgment module is used to perform multi-parameter joint evaluation and perform the second abnormality judgment; The steps of performing multi-parameter joint evaluation are: Get the power factor of smart meter box in real time; Obtain the total harmonic distortion rate of the smart meter box in real time; The total power supply and the total power of the user are obtained in real time, and the power balance error is obtained based on the total power supply and the total power of the user. The specific acquisition method is as follows: Where ΔE is the energy balance error, E 供 Expressed as the total power supply, E 用户 It is represented by the metered electric energy of each user, and M is the total number of users; Collect the instantaneous power of the smart meter box in real time, obtain the maximum instantaneous power and the minimum instantaneous power, and calculate the average power, the power load fluctuation coefficient and the power difference. The specific acquisition method is as follows: In the formula, λ represents the power load fluctuation coefficient, P(t) represents the instantaneous power, and P avg It is expressed as average power, max(P(t)) is expressed as maximum instantaneous power, min(P(t)) is expressed as minimum instantaneous power, P diff It is expressed as power variability; The power theft risk index is obtained based on the power factor, total harmonic distortion rate, power balance error and power load fluctuation coefficient. The specific method of obtaining it is as follows: In the formula, QD represents the risk index of electricity theft, λ represents the power load fluctuation coefficient, ΔE represents the power balance error, cosφ represents the power factor, THD represents the total harmonic distortion rate, and R diff It is expressed as power variability; The early warning module is used to issue early warning reminders for the second abnormal situation and remind relevant staff to conduct electricity theft inspection in time.

Images