Electric energy meter fault detection method based on waveform identification

Abstract

The invention relates to the technical field of electric energy meter fault detection, in particular to an electric energy meter fault detection method based on waveform recognition. The method comprises the following steps: acquiring a plurality of historical abnormal output waveforms and a current output waveform of a current electric energy meter; acquiring the fluctuation degree of each two adjacent historical wave bands in each historical abnormal output waveform; selecting the historical wave band with the maximum fluctuation degree as an abnormal wave band; obtaining a target wavelet coefficient waveform of the abnormal wave band under a corresponding target scale; dividing the current output waveform to obtain a plurality of waveform segments; calculating the matching degree between the wavelet coefficient waveform of any waveform segment and each target wavelet coefficient waveform corresponding to each abnormal waveband; and if the matching degree of at least one waveform segment is greater than the threshold value, determining that the current electric energy meter has a fault. The scheme of the invention can accurately identify the fault of the electric energy meter.

Description

Technical Field

[0001] The present invention relates to the technical field of power meter fault detection. More specifically, the present invention relates to a power meter fault detection method based on waveform recognition. Background Art

[0002] With the continuous increase in power demand, the importance of power meters in the power system has become increasingly prominent. It is not only used to record household and industrial electricity consumption, but also plays a key role in energy management and grid balancing.

[0003] Among them, the structure of the power meter includes a power metering chip, a communication module, a display screen, a power supply module, a collection and control unit, and a power meter housing, etc.; its working principle is to calculate the power consumption by measuring the voltage and current in the circuit. Specifically, when the current flows through the power metering chip, the chip will calculate the power consumption according to the magnitude of the current and voltage, and then transmit the data to the communication module and display it through the display screen.

[0004] Therefore, as an important metering device in the power system, the accuracy of the power meter is directly related to the normal operation of the power system and the user's electricity consumption experience. Power meter faults may lead to problems such as inaccurate metering, data loss, or abnormal fluctuations, so it is crucial to detect whether the power meter is faulty in a timely manner.

[0005] In the related art, for example, a Chinese patent application document with the publication number CN110596635A discloses a power meter post-fault detection method and a power meter. It analyzes the output waveform data, calculates the time-domain difference per cycle in the recorded waveform data by using the fast Fourier transform FFT, calculates the waveform characteristic values reflecting the waveform data, and then compares them with the fault judgment threshold to determine whether there is a power meter post-fault.

[0006] However, when performing fault analysis, the output waveform data may be affected by various factors, resulting in an increase in its non-stationarity, and the Fourier transform cannot effectively capture this non-stationarity. Therefore, the above solution cannot accurately detect the faults of the power meter. Summary of the Invention

[0007] The object of the present invention is to propose a power meter fault detection method based on waveform recognition to solve the problem that the faults of the power meter cannot be accurately detected in the prior art; for this purpose, the present invention provides a solution in the following aspect.

[0008] The power meter fault detection method based on waveform recognition provided by the present invention includes: Obtain a plurality of historical abnormal output waveforms and the current output waveform of the current power meter; Divide each historical abnormal output waveform to obtain a plurality of historical wave bands; Obtain the fluctuation degree between two adjacent historical bands in each historical abnormal output waveform, where the waveform degree characterizes the difference between the wavelet coefficients of two adjacent historical bands after wavelet transform; select the historical band with the largest fluctuation degree as the abnormal band; Obtain the variance of the wavelet coefficients of any abnormal band at different scales, select the scale when the variance is greater than the set threshold as the target scale, and then obtain the target wavelet coefficient waveform of the abnormal band at the corresponding target scale; Divide the current output waveform to obtain multiple waveform segments; Calculate the matching degree between the wavelet coefficient waveform of any waveform segment and the target wavelet coefficient waveforms corresponding to each abnormal band; the wavelet coefficient waveform is obtained by performing wavelet transform using the target scale of the target wavelet coefficient waveform during matching.

[0009] In the above solution, by obtaining the historical abnormal output waveform and analyzing the historical abnormal output waveform, the real abnormal historical band in the historical output waveform can be determined. Furthermore, the target scale affecting the abnormality of the historical band and the target wavelet coefficient waveform at the corresponding target scale can be obtained from the real abnormal historical band. When analyzing each waveform segment in the current output waveform, by matching the wavelet coefficients of each waveform segment and each target wavelet coefficient waveform at the same scale, the matching situation between the current waveform segment and the target wavelet coefficient waveform can be accurately evaluated, and thus the fault detection of the electric energy meter can be accurately performed.

[0010] Optionally, the fluctuation degree is: ; is the mean value of the wavelet variances of the m-th historical band in the i-th historical abnormal output waveform at all scales, is the mean value of the wavelet variances of the (m - 1)-th historical band in the i-th historical abnormal output waveform at all scales, and M is the total number of historical bands in the i-th historical abnormal output waveform; the wavelet variance is the variance of all wavelet coefficients of each historical band at the same scale.

[0011] In the above solution, by obtaining the mean value of the wavelet variances of each historical band in the historical abnormal output waveform at all scales, the fluctuation situation of each historical band at all scales can be characterized.

[0012] Optionally, the fluctuation degree is ; where, is the mean value of all wavelet coefficients of the m-th historical band in the i-th historical abnormal output waveform at all scales, is the mean of all wavelet coefficients of the (m - 1)-th historical band in the i-th historical abnormal output waveform at all scales, and M is the total number of historical bands in the i-th historical abnormal output waveform.

[0013] Optionally, before determining the target scale, it is also necessary to screen each abnormal band to obtain the target band, and the specific steps are as follows: Obtain the load of each abnormal band, and screen out the abnormal bands whose difference from the load of the current output waveform is less than the difference threshold as the target bands.

[0014] The above solution can screen out the target bands with the same or similar load as the current output waveform for subsequent matching, improving the matching accuracy and reducing the calculation amount.

[0015] Optionally, taking any one of the historical abnormal output waveforms and the current output waveform as the target waveform, the process of dividing each target waveform is as follows: Obtain the extreme points in each target waveform; Determine the peak width and peak height of the local region where each extreme point is located; Calculate the absolute value of the difference in peak width and the absolute value of the difference in peak height between any extreme point and its adjacent extreme point. If the absolute value of the difference in peak width is less than or equal to the first threshold, and the absolute value of the difference in peak height is less than or equal to the second threshold, then the two extreme points are divided into one waveform segment, and so on, traversing all extreme points to achieve the division of the target waveform; otherwise, the two extreme points are located in different waveform segments.

[0016] The above solution accurately divides the target waveform and classifies similar or identical maximum points into one category.

[0017] Optionally, the matching degree is obtained by DTW.

[0018] The above solution can obtain an accurate matching degree.

[0019] Optionally, the process of obtaining multiple historical abnormal output waveforms is as follows: Obtain multiple historical output waveforms, label each historical output waveform, and obtain the historical abnormal output waveforms.

[0020] The above solution can directly obtain the historical abnormal output waveforms.

[0021] Optionally, the historical output waveform is the waveform data output by the analog-to-digital converter of the electric energy metering chip in the electric energy meter; the current output waveform is the waveform data output by the analog-to-digital converter of the current electric energy meter, and the waveform data is a voltage waveform or a current waveform.

[0022] Optionally, the method further includes: performing denoising processing on the historical abnormal output waveform and the current output waveform respectively to obtain the denoised historical abnormal output waveform and the current output waveform.

[0023] Optionally, the method further includes: sending the acquired fault condition of the current electric energy meter to a staff member to remind the staff member to perform maintenance.

[0024] The beneficial effects of the present invention are: The solution of the present invention performs wavelet decomposition and fluctuation degree analysis on the historical abnormal output waveform, fully utilizes the local characteristics of the historical abnormal output waveform, and realizes fault detection of the output waveform of the current electricity meter through matching, thereby solving the problem that traditional methods are difficult to accurately capture non-stationary waveform characteristics, and has the advantage of improving fault detection accuracy and sensitivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The following is a schematic flowchart showing the steps of the electric energy meter fault detection method based on waveform recognition in this embodiment; Figure 2 The flowchart schematically shows the steps of obtaining the target wavelet coefficient waveform in the electric energy meter fault detection method based on waveform recognition in this embodiment. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0027] In power systems, the proper operation of energy meters is crucial. A fault in an energy meter can lead to inaccurate power metering, impacting the normal operation of the power system and the user's electricity experience. To improve the accuracy of energy meter fault detection, this paper proposes an energy meter fault detection method based on waveform recognition. By analyzing multiple historical output waveforms and the current output waveform of the current energy meter, accurate detection of the current energy meter fault can be achieved.

[0028] Taking an electronic electric energy meter as an example, the electric energy meter fault detection method based on waveform recognition of the present invention is introduced.

[0029] Specifically, if Figure 1 As shown, the electric energy meter fault detection method based on waveform recognition in this embodiment includes the following steps: Step S1, obtaining the current output waveform of the current electric energy meter.

[0030] In this embodiment, the current output waveform collected is the waveform data output by the analog-to-digital converter of the electric energy metering chip in the current electric energy meter, wherein the waveform data is a voltage waveform and / or a current waveform.

[0031] Among them, the analog-to-digital converter of the electric energy metering chip can convert analog signals into digital signals and output accurate waveform data.

[0032] Furthermore, the current output waveform is further denoised to obtain the denoised current output waveform. Among them, the denoising process can use Kalman filtering. Since Kalman filtering is a prior art, it will not be elaborated here.

[0033] The above denoising process can eliminate or reduce the noise interference introduced into the waveform data due to external interference or other factors, improving the accuracy and reliability of the waveform data. Through the denoised waveform data, the quality of the waveform data is ensured, and subsequent waveform matching and fault detection can be carried out more accurately, thereby improving the accuracy and reliability of the electric energy meter fault detection.

[0034] Step S2, obtain the target wavelet coefficient waveform.

[0035] The process of obtaining the target wavelet coefficient waveform includes steps S21 - S25, specifically as follows: Step S21, obtain multiple historical output waveforms.

[0036] Among them, the historical output waveform is the waveform data output by the analog-to-digital converter of the electric energy metering chip in the electric energy meter. The historical output waveform includes a voltage waveform and / or a current waveform.

[0037] In practical applications, the voltage waveform and current waveform of the electric energy meter can be collected regularly and stored in the database as historical output waveforms. When fault detection is required, the current voltage waveform and current waveform are collected in real time through the analog-to-digital converter of the electric energy metering chip as the current output waveform. Then, the current output waveform is compared and analyzed with the historical output waveform to identify whether there is a fault in the electric energy meter. The data consistency between the historical output waveform and the current output waveform ensures the comparability and reliability of the basic data in the fault detection process.

[0038] Of course, as other implementation manners, the historical output waveform can also be the historical waveform data of other same-type electric energy meters.

[0039] Step S22, label each historical output waveform to obtain a historical abnormal output waveform. Specifically, according to the operating conditions of the electric energy meter corresponding to the obtained historical output waveform, each historical output waveform can be labeled as to whether there is a fault.

[0040] Step S23, divide each historical abnormal output waveform to obtain multiple historical wave bands.

[0041] In this embodiment, the process of dividing each historical abnormal output waveform is as follows: Obtain the extreme points in each historical abnormal output waveform; determine the peak width and peak height of the local region where each extreme point is located; calculate the absolute value of the difference in peak width and the absolute value of the difference in peak height between any extreme point and its adjacent extreme point. If the absolute value of the difference in peak width is less than or equal to the first threshold and the absolute value of the difference in peak height is less than or equal to the second threshold, then divide the two extreme points into one waveform segment. By analogy, traverse all extreme points to achieve the division of the historical abnormal output waveform; otherwise, the two extreme points are located in different waveform segments.

[0042] The extreme points in the above historical abnormal output waveform can be realized by analyzing the local maximum and minimum values of the waveform data. Specifically, use the sliding window method to traverse the waveform data to determine the positions of the local extreme points.

[0043] The peak width and peak height of the local region can be determined by calculating the data change within a set range around the extreme point. Specifically, the peak width can be defined as the distance between the two abscissas corresponding to the preset threshold after the values on both sides of the extreme point drop to the preset threshold; the peak height can be defined as the height difference between the extreme point and the lowest point around it. Among them, the preset threshold can be a value randomly selected on the waveform that is less than the ordinate of the extreme point.

[0044] By obtaining the extreme points in the historical abnormal output waveform and determining the peak width and peak height of the local region where each extreme point is located, the characteristics of the waveform can be effectively identified.

[0045] At the same time, calculate the absolute value of the difference in peak width and peak height between extreme points, and compare these absolute values of differences with the corresponding first threshold and second threshold, so as to divide similar extreme points into the same wave band. That is, this embodiment can achieve precise division of the historical abnormal output waveform, which is helpful for subsequent fault detection.

[0046] As a preferred implementation manner, the first threshold and the second threshold can be determined according to empirical data in actual applications or through experiments. The selection of the first threshold and the second threshold should ensure that different types of waveform characteristics can be effectively distinguished to achieve precise waveform division.

[0047] Step S24, select the historical wave band with the largest fluctuation degree as the abnormal wave band.

[0048] In this embodiment, the process of obtaining the abnormal wave band is as follows: First, perform wavelet decomposition on each historical wave band to obtain wavelet coefficients at different scales and positions. Specifically, use discrete wavelet transform to perform wavelet decomposition on each historical wave band.

[0049] Second, obtain the fluctuation degrees of adjacent historical wave bands in each historical abnormal output waveform.

[0050] In one embodiment, the degree of fluctuation is ; is the mean value of the wavelet variances of the m-th historical wave band in the i-th historical abnormal output waveform at all scales, is the mean value of the wavelet variances of the (m - 1)-th historical wave band in the i-th historical abnormal output waveform at all scales, and M is the total number of historical wave bands in the i-th historical abnormal output waveform.

[0051] Among them, the wavelet variance is obtained according to the wavelet coefficients at different positions of the corresponding historical wave band at the same scale. Specifically, the variance of the wavelet coefficients can be obtained by performing statistical analysis on the wavelet coefficients at different positions at the same scale. By calculating the wavelet variance, the change characteristics of the waveform data at different positions can be captured.

[0052] As another implementation manner, the mean value of the above wavelet variances can also be a weighted variance, that is, obtained by performing weighted averaging on the wavelet variances. The weights in the weighting process can be determined according to the actual situation.

[0053] In another embodiment, the degree of fluctuation is ; Among them, is the mean value of all wavelet coefficients of the m-th historical wave band in the i-th historical abnormal output waveform at all scales, is the mean value of all wavelet coefficients of the (m - 1)-th historical wave band in the i-th historical abnormal output waveform at all scales, and M is the total number of historical wave bands in the i-th historical abnormal output waveform.

[0054] The above embodiment characterizes the degree of fluctuation by calculating the difference in the mean values of the wavelet coefficients of two adjacent historical wave bands in the historical abnormal output waveform, and can capture the abnormal fluctuation segments in the historical abnormal output waveform.

[0055] Then, select the historical wave band with the largest degree of fluctuation as the abnormal wave band.

[0056] The acquisition of the above degree of fluctuation can more accurately capture the abnormal fluctuations in the historical abnormal output waveform, providing data support for subsequent fault detection of the electric energy meter.

[0057] Step S25: Obtain the variance of the wavelet coefficients of any abnormal wave band at different scales, select the scale when the variance is greater than the set threshold as the target scale, and then obtain the target wavelet coefficient waveform of the abnormal wave band at the corresponding target scale.

[0058] In this embodiment, by performing wavelet transform on each abnormal band, wavelet coefficients at different positions under the same scale can be obtained, and further wavelet coefficients at different positions under all scales can be obtained.

[0059] In this embodiment, the variances of the wavelet coefficients at different positions of each abnormal band under one scale are calculated, and these variances are used as the characteristic quantities of the abnormal bands under the corresponding scales.

[0060] It should be noted that one abnormal band corresponds to multiple scales, and the wavelet coefficients at all positions under each scale form a wavelet coefficient waveform.

[0061] After obtaining the characteristic quantities of any abnormal band under different scales, in this embodiment, the characteristic quantities also need to be evaluated to determine the scale corresponding to the actual abnormality of the any abnormal band. Specifically, when the characteristic quantity (variance) is greater than the set threshold, the corresponding scale is the target scale of the any abnormal band.

[0062] The wavelet coefficient waveform under the above target scale can reflect the abnormal situation of the abnormal band. Among them, the target scale of any abnormal band can be multiple or one, which is determined by the abnormal situation of the abnormal band.

[0063] The above set threshold is 0.1. It can be specifically determined according to the actual situation. Of course, it can also be the mean value of the variances of the wavelet coefficients of the historical normal output waveforms of the watt-hour meter under all scales when it is normal.

[0064] Further, before determining the target scale, it is also necessary to screen each abnormal band to obtain the target band. The specific steps are as follows: Obtain the loads of each abnormal band, and screen out the abnormal bands whose differences from the load of the current output waveform are less than the difference threshold as the target bands.

[0065] In this embodiment, the loads of each abnormal band and the load of the current output waveform are obtained, and the absolute value of the difference between the two loads is compared with the difference threshold to determine the abnormal bands that are the same or similar to the current output waveform, so as to improve the accuracy of subsequent calculations.

[0066] The above load size is represented by power; the value of the above difference threshold is 1 watt. Of course, it can also be determined according to the actual situation.

[0067] Step S3: Divide the current output waveform to obtain multiple waveform segments; calculate the matching degrees between the wavelet coefficient waveform of any waveform segment and the target wavelet coefficient waveforms corresponding to each abnormal band. If the matching degree of at least one waveform segment is greater than the threshold, the current watt-hour meter has a fault.

[0068] The method for dividing the current output waveform is the same as that for dividing each historical abnormal output waveform, which will not be elaborated here.

[0069] When calculating the matching degree, in this embodiment, multiple waveform segments are subjected to wavelet transform to obtain the corresponding wavelet coefficient waveforms, and then the matching between the wavelet coefficient waveforms and the corresponding target wavelet coefficient waveforms is performed. It should be noted that for the wavelet transform of each waveform segment, it is carried out according to the target scale corresponding to the target wavelet coefficient waveform during matching.

[0070] When the above-mentioned waveform segments are matched, the matching of the wavelet coefficient waveforms under the same scale parameter can improve the accuracy of matching.

[0071] Exemplarily, taking the fluctuation segment A of the current output waveform as an example, any abnormal waveband corresponds to the target wavelet coefficient waveforms B, C, and D under three target scales. Then when the fluctuation segment A is matched with the target wavelet coefficient waveform B, it is necessary to first perform wavelet transform on the fluctuation segment A using the target scale corresponding to the target wavelet coefficient waveform B to obtain the wavelet coefficient waveform, and calculate the matching degree between the wavelet coefficient waveform and the target wavelet coefficient waveform B; for the matching between the fluctuation segment A and the target wavelet coefficient waveform C, it is still necessary to first perform wavelet transform on the fluctuation segment A using the target scale corresponding to the target wavelet coefficient waveform C to obtain the corresponding wavelet coefficient waveform, and then match it with the target wavelet coefficient waveform C.

[0072] The above-mentioned matching degree is calculated by using the dynamic time warping (DTW) algorithm to calculate the similarity between the current waveform segment and the target wavelet coefficient waveforms of each abnormal waveband.

[0073] Among them, the magnitude of the matching degree value reflects the similarity between the current output waveform of the current electric energy meter and the historical abnormal output waveform.

[0074] The DTW algorithm, that is, the dynamic time warping algorithm, is an algorithm used to measure the similarity between two time series. By using the DTW algorithm, the similarity between the current waveform segment and the target wavelet coefficient waveforms under the target scale in each abnormal waveband can be effectively compared, so as to accurately judge whether there is a fault in the electric energy meter. When processing time series data, the DTW algorithm has the advantages of being able to handle different sequence lengths and local offsets on the time axis, so it has high accuracy and robustness in waveform matching.

[0075] By using the DTW algorithm for waveform matching, the present invention can more accurately judge whether there is a fault in the electric energy meter, improving the accuracy and reliability of fault detection. Compared with the prior art, the method in this embodiment can handle the problems of different time series lengths and local offsets on the time axis, and has higher accuracy and robustness.

[0076] Further, when there is no case where the matching degree is greater than the threshold value, the current electricity meter has no fault.

[0077] Among them, the value of the threshold is 0.95; of course, it can also be determined according to the actual situation.

[0078] In this embodiment, by setting the conditions of the matching degree and the threshold value, misjudgment is effectively avoided, and the accuracy of the electricity meter fault detection is ensured.

[0079] Further, in this embodiment, the fault type and fault location of the transmission line are sent to relevant personnel to remind the relevant personnel to carry the equipment related to the fault and perform maintenance in time.

[0080] The solution of the present invention can more accurately judge the fault state of the electricity meter, avoid misjudgment caused by the non-stationarity of the waveform data, and improve the reliability of the electricity meter fault detection.

[0081] In the description of this specification, the meaning of "a plurality" is at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

Claims

1. A method for detecting faults in an electric energy meter based on waveform recognition, characterized in that, Including: Obtain multiple historical abnormal output waveforms and the current output waveform of the current electricity meter; Divide each historical abnormal output waveform to obtain multiple historical wave bands; Obtain the fluctuation degree between two adjacent historical wave bands in each historical abnormal output waveform, where the waveform degree characterizes the difference between the wavelet coefficients of two adjacent historical wave bands after wavelet transform; select the historical wave band when the fluctuation degree is the largest as the abnormal wave band; Obtain the variance of the wavelet coefficients of any abnormal wave band at different scales, select the scale when the variance is greater than the set threshold as the target scale, and then obtain the target wavelet coefficient waveform of the abnormal wave band at the corresponding target scale; Divide the current output waveform to obtain multiple waveform segments; Calculate the matching degree between the wavelet coefficient waveform of any waveform segment and the target wavelet coefficient waveforms corresponding to each abnormal wave band; the wavelet coefficient waveform is obtained by performing wavelet transform using the target scale of the target wavelet coefficient waveform during matching; If the matching degree of at least one waveform segment is greater than the threshold, the current electricity meter has a fault.

2. The method for detecting faults of an electric energy meter based on waveform recognition according to claim 1, wherein The degree of fluctuation is as follows: ； is the mean value of the wavelet variances of the m-th historical wave band in the i-th historical abnormal output waveform at all scales, is the mean value of the wavelet variances of the (m - 1)-th historical wave band in the i-th historical abnormal output waveform at all scales, where M is the total number of historical wave bands in the i-th historical abnormal output waveform; the wavelet variance is the variance of all wavelet coefficients of each historical wave band at the same scale.

3. The method for detecting the failure of an electric energy meter based on waveform recognition according to claim 1, wherein, The degree of fluctuation is ； Among them, is the mean value of all wavelet coefficients of the m-th historical band in the i-th historical abnormal output waveform at all scales, is the mean value of all wavelet coefficients of the (m - 1)-th historical band in the i-th historical abnormal output waveform at all scales, and M is the total number of historical bands in the i-th historical abnormal output waveform.

4. The method for detecting faults of an electric energy meter based on waveform recognition according to claim 1, characterized in that, Before determining the target scale, it is also necessary to screen each abnormal wave band to obtain the target wave band. The specific steps are: Obtain the load of each abnormal wave band, and screen out the abnormal wave bands whose difference from the load of the current output waveform is less than the difference threshold as the target wave bands.

5. The method for detecting the failure of an electric energy meter based on waveform recognition according to claim 1, wherein Taking any waveform among the multiple historical abnormal output waveforms and the current output waveform as the target waveform, the process of dividing each target waveform is: Obtain the extreme points in each target waveform; Determine the peak width and peak height of the local region where each extreme point is located; Calculate the absolute value of the difference in peak width and the absolute value of the difference in peak height between any extreme point and its adjacent extreme point. If the absolute value of the difference in peak width is less than or equal to the first threshold and the absolute value of the difference in peak height is less than or equal to the second threshold, then the two extreme points are divided into one waveform segment. By analogy, traverse all extreme points to achieve the division of the target waveform; otherwise, the two extreme points are located in different waveform segments.

6. The method for detecting the failure of an electric energy meter based on waveform recognition according to claim 1, wherein The matching degree is obtained using the DTW algorithm.

7. The method for detecting the failure of an electric energy meter based on waveform recognition according to claim 2 or 3, characterized in that, The process of obtaining multiple historical abnormal output waveforms is: Obtain multiple historical output waveforms, and label each historical output waveform to obtain historical abnormal output waveforms.

8. The method for detecting the failure of an electric energy meter based on waveform recognition according to claim 7, wherein, The historical output waveform is the waveform data output by the analog-to-digital converter of the electric energy metering chip in the electricity meter; the current output waveform is the waveform data output by the analog-to-digital converter of the current electricity meter, and the waveform data is a voltage waveform or a current waveform.

9. The method for detecting the failure of an electric energy meter based on waveform recognition according to claim 1, characterized in that, Also including: Perform denoising processing on the historical abnormal output waveform and the current output waveform respectively to obtain the denoised historical abnormal output waveform and the current output waveform.

10. The method for detecting the failure of an electric energy meter based on waveform recognition according to claim 1, wherein, Also including: Send the obtained fault situation of the current electricity meter to the staff to remind the staff to perform maintenance.

Images