Electric energy meter fault detection method and device, terminal equipment and computer readable storage medium

Abstract

The invention discloses an electric energy meter fault detection method and device, terminal equipment and a computer readable storage medium, and the method comprises the steps: obtaining the voltage data, current data and temperature and humidity data of an electric energy meter in a current sampling time period, obtaining the historical feedback layer gain of the electric energy meter in a previous sampling time period, and determining a final gain parameter; multiplying the final gain parameter by the current bandwidth and the current cut-off frequency of the anti-interference filter to obtain the updated bandwidth and cut-off frequency of the anti-interference filter; when the anti-interference filter works based on the updated bandwidth and cut-off frequency, acquiring a voltage fluctuation value and a current fluctuation value filtered by the anti-interference filter; and comparing the voltage fluctuation value with a voltage fluctuation threshold value, comparing the current fluctuation value with a current fluctuation threshold value, if the voltage fluctuation value is greater than the voltage fluctuation threshold value and the current fluctuation value is greater than the current fluctuation threshold value, judging that the electric energy meter has a high-interference fault, otherwise, judging that the electric energy meter does not have the high-interference fault.

Description

Technical Field

[0001] The present invention relates to the field of power meter fault detection, and particularly to a power meter fault detection method, device, terminal device and computer-readable storage medium. Background Art

[0002] With the rapid development of the smart grid and the wide application of power meters in the field of power monitoring, as a key power consumption metering device in the power system, the operation stability and measurement accuracy of power meters directly affect the data reliability of the power system.

[0003] However, the traditional detection methods of power meters mainly rely on a single threshold detection method. When facing interference signals with extremely high diversity and uncertainty, the traditional threshold detection method is extremely prone to false alarms or missed alarms due to the diversity and uncertainty of interference signals, resulting in inaccurate fault identification. Correspondingly, in practical applications, it is very difficult to find a threshold that neither generates false alarms due to interference signals nor can accurately detect fault signals. If the threshold is set too low, the system may generate false alarms for normal fluctuations or noises; if the threshold is set too high, some real fault signals may be missed. Summary of the Invention

[0004] Embodiments of the present invention provide a power meter fault detection method, device, terminal device and computer-readable storage medium, which can improve the reliability of power meter fault identification in a complex interference environment.

[0005] An embodiment of the present invention provides a power meter fault detection method, including:

[0006] Obtain the voltage data, current data and temperature and humidity data of the power meter in the current sampling period, and the historical feedback layer gain in the previous sampling period;

[0007] Determine the final gain parameter according to the voltage data, current data and temperature and humidity data of the current sampling period, and the historical feedback layer gain in the previous sampling period;

[0008] Multiply the final gain parameter by the current bandwidth and the current cut-off frequency of the anti-interference filter respectively to obtain the updated bandwidth and the updated cut-off frequency of the anti-interference filter;

[0009] When the anti-interference filter operates based on the updated bandwidth and the updated cut-off frequency, obtain the voltage fluctuation value and the current fluctuation value filtered by the anti-interference filter;

[0010] Compare the voltage fluctuation value with the voltage fluctuation threshold, and compare the current fluctuation value with the current fluctuation threshold. If the voltage fluctuation value is greater than the voltage fluctuation threshold and the current fluctuation value is greater than the current fluctuation threshold, it is determined that the watt-hour meter has a high interference fault; otherwise, it is determined that the watt-hour meter does not have a high interference fault.

[0011] Further, based on the voltage data, current data, temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period, determine the final gain parameter, including:

[0012] Based on the voltage data, current data, temperature and humidity data of the current sampling period, calculate the voltage fluctuation amplitude, current fluctuation amplitude, temperature factor and humidity factor;

[0013] Perform a fast Fourier transform on the voltage data, current data, temperature and humidity data, and select the frequency component with the highest spectral density as the environmental interference frequency;

[0014] Divide the sum of the voltage fluctuation amplitude and the current fluctuation amplitude by the environmental interference frequency to obtain the interference intensity factor;

[0015] Determine the interference layer gain according to the interference intensity factor;

[0016] Sum and average the temperature factor and the humidity factor to obtain the temperature and humidity layer gain;

[0017] Based on the historical feedback layer gain of the previous sampling period, calculate the historical gain change rate and the historical feedback layer gain of the current sampling period;

[0018] Based on the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain, and historical feedback layer gain of the current sampling period, determine the final gain parameter.

[0019] Further, based on the historical feedback layer gain of the previous sampling period, calculate the historical gain change rate and the historical feedback layer gain of the current sampling period, including:

[0020] Based on the historical feedback layer gain of the previous sampling period, calculate the historical gain change rate and the historical feedback layer gain of the current sampling period through the following formula:

[0021]

[0022] G h =1+|R G |;

[0023] where, R G represents the historical gain change rate of the current sampling period, G′ h (t - T wrepresents the historical feedback layer gain of the previous sampling period, G′ h (t - 2·T w ) represents the historical feedback layer gain of the previous two sampling periods, t represents the current sampling time, T w represents the sampling window length, G h represents the historical feedback layer gain of the current sampling period.

[0024] Furthermore, based on the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain, and historical feedback layer gain of the current sampling period, determine the final gain parameter, including:

[0025] Based on the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain, and historical feedback layer gain of the current sampling period, calculate the final gain parameter through the following formula:

[0026]

[0027] G(t) = W d ·G d +W TH ·G TH +W h ·G h ;

[0028] where, W d represents the interference layer weight of the current sampling period, S d represents the interference intensity factor of the current sampling period, K T represents the temperature factor of the current sampling period, K H represents the humidity factor of the current sampling period, W TH represents the temperature and humidity layer weight of the current sampling period, W h represents the historical feedback layer weight of the current sampling period, G d represents the interference layer gain of the current sampling period, G TH represents the temperature and humidity layer gain of the current sampling period, G(t) represents the final gain parameter.

[0029] Furthermore, after determining that the electricity meter has a high interference fault, it also includes:

[0030] Take the electricity meter determined to have a high interference fault as the target electricity meter;

[0031] Establish a multivariate Gaussian distribution model based on the voltage data of the target electricity meter;

[0032] Based on the multivariate Gaussian distribution model, calculate the log-likelihood value of the voltage data of the target electricity meter;

[0033] Compare the log-likelihood value of the target electricity meter voltage data with a preset voltage threshold;

[0034] If the log-likelihood value of the target electricity meter voltage data is greater than or equal to the preset voltage threshold, it is determined that the voltage measurement module in the target electricity meter is faulty;

[0035] If the log-likelihood value of the target electricity meter voltage data is less than the preset voltage threshold, determine the fault type and probability of the target electricity meter through a fault identification model.

[0036] Further, determining the fault type and probability of the target electricity meter through a fault identification model includes:

[0037] Perform weighted processing on the fault characteristics of the target electricity meter according to a preset optimal weight combination to obtain weighted fault characteristics; the fault characteristics include: voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate, and final gain parameter in the current sampling period;

[0038] Input the weighted fault characteristics into a trained fault type identification model so that the fault identification model generates each fault type of the target electricity meter and its corresponding probability according to the weighted fault characteristics.

[0039] Further, the preset optimal weight combination is determined by the following method:

[0040] Obtain a number of training samples; each training sample includes: historical fault characteristics of the electricity meter and its corresponding actual faults; the historical fault characteristics include: voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate, and final gain parameter in the historical sampling period;

[0041] Determine the number of weight combinations according to a preset population size;

[0042] Randomly initialize each weight combination to obtain a number of initialized weight combinations; where the weight combination includes the weight value of each feature in the historical fault characteristics;

[0043] Repeat the wild horse optimization process until the preset maximum number of iterations is reached, or the loss value of the leading combination converges, to obtain the optimal weight combination;

[0044] The wild horse optimization process includes:

[0045] For each current weight combination, perform weighted processing on the current weight combination and the historical fault characteristics in each training sample to obtain the weighted historical fault characteristics of each training sample; where the current weight combination is the initialized weight combination when the wild horse optimization process is executed for the first time;

[0046] Input the weighted historical fault features of each training sample into the fault type recognition model to be trained, so that the fault type recognition model to be trained takes the weighted historical fault features of each training sample as input and outputs the predicted fault type of each training sample and its corresponding probability. During the training process, calculate the cross-entropy loss function according to the probability of the predicted fault type of each training sample and the actual fault of the corresponding training sample, and adjust the network parameters of the fault type recognition model according to the cross-entropy loss function until the cross-entropy loss function converges to obtain a pre-trained fault type recognition model;

[0047] Take the convergence value of the cross-entropy loss function as the loss value;

[0048] Take the current weight combination with the smallest loss value among all current weight combinations as the leading combination;

[0049] For each current weight combination, update the current weight combination according to the leading combination to obtain an updated weight combination;

[0050] Determine whether the preset maximum number of iterations is satisfied during this wild horse optimization process, or whether the loss value of the leading combination converges. If so, take the leading combination as the optimal weight combination; otherwise, take each updated weight combination as each current weight combination for the next execution of the wild horse optimization process.

[0051] Based on the above method item embodiments, the present invention correspondingly provides device item embodiments, including: a sampling data acquisition module, a gain parameter calculation module, a filter parameter update module, a fluctuation value acquisition module, and a high-interference fault discrimination module;

[0052] The sampling data acquisition module is used to acquire voltage data, current data, and temperature and humidity data of the current sampling period of the electric energy meter, as well as the historical feedback layer gain of the previous sampling period;

[0053] The gain parameter calculation module is used to determine the final gain parameter according to the voltage data, current data, and temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period;

[0054] The filter parameter update module is used to multiply the final gain parameter by the current bandwidth and the current cut-off frequency of the anti-interference filter respectively to obtain the updated bandwidth and the updated cut-off frequency of the anti-interference filter;

[0055] The fluctuation value acquisition module is used to acquire the voltage fluctuation value and the current fluctuation value after being filtered by the anti-interference filter when the anti-interference filter works based on the updated bandwidth and the updated cut-off frequency;

[0056] A high-interference fault discrimination module is used to compare the voltage fluctuation value with the voltage fluctuation threshold and compare the current fluctuation value with the current fluctuation threshold. If the voltage fluctuation value is greater than the voltage fluctuation threshold and the current fluctuation value is greater than the current fluctuation threshold, it is determined that the electric energy meter has a high-interference fault; otherwise, it is determined that the electric energy meter does not have a high-interference fault.

[0057] Based on the above method item embodiments, the present invention correspondingly provides terminal device item embodiments, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps of the electric energy meter fault detection method as described in the present invention are implemented.

[0058] Based on the above method item embodiments, the present invention correspondingly provides computer-readable storage medium item embodiments, including: a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the steps of the electric energy meter fault detection method as described in the present invention.

[0059] Compared with the prior art, the beneficial effects of the embodiments of the present solution are as follows:

[0060] The present invention obtains the voltage data, current data, and temperature and humidity data of the current sampling period of the electric energy meter, as well as the historical feedback layer gain of the previous sampling period. Through these parameters, the actual working environment of the electric energy meter can be more comprehensively understood. Among them, the voltage data and current data are the direct basis for judging whether there is interference. At the same time, considering the temperature and humidity data and historical feedback to determine the final gain parameter, so that the electric energy meter can adjust its final gain parameter according to different working environments and interference situations. Then, multiply the final gain parameter by the current bandwidth and current cut-off frequency of the anti-interference filter respectively to obtain the updated bandwidth and updated cut-off frequency of the anti-interference filter, so that the anti-interference filter can filter out the influence of environmental factors and external interference signals, thereby improving the measurement accuracy of the electric energy meter in a complex interference environment. When the anti-interference filter works based on the updated bandwidth and updated cut-off frequency, obtain the voltage fluctuation value and current fluctuation value filtered by the anti-interference filter, compare the voltage fluctuation value with the voltage fluctuation threshold, and compare the current fluctuation value with the current fluctuation threshold. Considering both the voltage fluctuation value and the current fluctuation value at the same time, only when both exceed their respective thresholds, it is determined that there is a high-interference fault, avoiding the false alarm or missed alarm problems that may be caused by single-parameter judgment and improving the reliability of fault identification. Description of the Drawings

[0061] Figure 1 It is a schematic flowchart of the electric energy meter fault detection method provided by an embodiment of the present invention;

[0062] Figure 2It is a schematic flow chart for determining the final gain parameter provided by an embodiment of the present invention;

[0063] Figure 3 It is a schematic structural diagram of an electric energy meter fault detection device provided by an embodiment of the present invention. Detailed implementation manners

[0064] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0065] As Figure 1 shown, an embodiment of the present invention provides an electric energy meter fault detection method, which at least includes the following steps:

[0066] Step S1: Obtain the voltage data, current data, temperature and humidity data of the current sampling period of the electric energy meter, and the historical feedback layer gain of the previous sampling period;

[0067] For step S1, based on the sensor devices configured in the electric energy meter, the voltage data and current signals are collected, and based on the temperature and humidity sensors configured in the electric energy meter, the temperature and humidity data are collected, where the temperature and humidity data include temperature data and humidity data.

[0068] Obtain the historical feedback layer gain of the previous sampling period from the database for subsequent comparison and analysis with the data of the current sampling period. It should be noted that if the electric energy meter has just started running and there is no historical feedback layer gain for the previous sampling period, a preset default value is used as the historical feedback layer gain for the previous sampling period.

[0069] Step S2: Determine the final gain parameter according to the voltage data, current data, temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period;

[0070] For step S2, considering the voltage data, current data, temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period, multiple data sources are fused and analyzed to obtain a gain parameter that can accurately reflect the current state of the system and adapt to future changes.

[0071] Specifically, as Figure 2 shown, determining the final gain parameter according to the voltage data, current data, temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period includes the following steps:

[0072] Step S21: Calculate the voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, and humidity factor based on the voltage data, current data, and temperature and humidity data in the current sampling period;

[0073] For step S21, calculate the voltage fluctuation amplitude and current fluctuation amplitude respectively based on the difference between the maximum and minimum values of the voltage data and current data within the sampling window.

[0074] Based on the temperature data and humidity data, calculate the normalized deviation ratio in combination with the reference temperature value and reference humidity value, and use them as the temperature factor and humidity factor respectively.

[0075] By calculating the voltage and current fluctuation amplitudes respectively based on the difference between the maximum and minimum values of the voltage data and current signal within the sampling window, the instantaneous fluctuations of voltage and current during the operation of the watt-hour meter can be identified more accurately. Calculating the temperature factor and humidity factor based on the normalized deviation ratio of the temperature data and humidity data helps to quantify the impact of environmental factors on the operation of the watt-hour meter, enabling the watt-hour meter to maintain measurement accuracy under different environmental conditions.

[0076] Step S22: Perform a fast Fourier transform on the voltage data, current data, and temperature and humidity data, and select the frequency component with the highest spectral density as the environmental interference frequency;

[0077] For step S22, perform a fast Fourier transform (FFT) on the voltage data and current data within the acquisition window to convert the time-domain signal into a frequency-domain signal, and select the frequency component with the highest spectral density from the results of the fast Fourier transform (FFT) as the environmental interference frequency of the current window.

[0078] Step S23: Divide the sum of the voltage fluctuation amplitude and current fluctuation amplitude by the environmental interference frequency to obtain the interference intensity factor;

[0079] For step S23, based on the fluctuation amplitude and environmental interference frequency within the sampling window, calculate the interference intensity factor through the following formula:

[0080]

[0081] where the fluctuation amplitude includes the voltage fluctuation amplitude and current fluctuation amplitude, S d represents the interference intensity factor of the current sampling period, ΔV represents the voltage fluctuation amplitude of the current sampling period, ΔI represents the current fluctuation amplitude of the current sampling period, and f d represents the environmental interference frequency.

[0082] Step S24: Determine the interference layer gain according to the interference intensity factor;

[0083] For step S24, the sliding mode control technology is introduced to define the interference layer gain, so that the gain is adaptively adjusted according to the interference intensity. Specifically, the interference layer gain is determined by the following formula:

[0084]

[0085] where G d represents the interference layer gain in the current sampling period. In the form of a square root function, the interference layer gain is more sensitive to smaller interferences and gradually stabilizes under larger interferences, avoiding instability caused by excessive gain.

[0086] Step S25: Sum and average the temperature factor and the humidity factor to obtain the temperature and humidity layer gain;

[0087] For step S25, the temperature factor and the humidity factor are introduced, and the temperature and humidity layer gain is defined by an exponential adaptive function. Specifically, the temperature factor and the humidity factor are summed and averaged to obtain the temperature and humidity layer gain, which is mathematically expressed as follows:

[0088]

[0089] where G TH represents the temperature and humidity layer gain in the current sampling period, represents the temperature factor, represents the humidity factor.

[0090] Step S26: Calculate the historical gain change rate and the historical feedback layer gain in the current sampling period based on the historical feedback layer gain in the previous sampling period;

[0091] In a preferred embodiment, calculating the historical gain change rate and the historical feedback layer gain in the current sampling period based on the historical feedback layer gain in the previous sampling period includes:

[0092] Calculate the historical gain change rate and the historical feedback layer gain in the current sampling period based on the historical feedback layer gain in the previous sampling period through the following formula:

[0093]

[0094] G h = 1 + |R G |;

[0095] where R G represents the historical gain change rate in the current sampling period, G′ h (t - T w ) represents the historical feedback layer gain in the previous sampling period, G′ h (t - 2·T w) represents the historical feedback layer gain for the first two sampling periods, t represents the current sampling time, and T w represents the sampling window length, and G h represents the historical feedback layer gain for the current sampling period.

[0096] For step S26, the historical gain change rate is obtained by comparing the gain of the previous sampling period with that of the sampling period before the previous one for the current sampling period, and is used to measure the degree of change in the gain between two consecutive sampling periods. After calculating the historical gain change rate, an operation of taking the absolute value of the historical gain change rate and adding 1 is performed to ensure that the historical feedback layer gain is always positive and reflects the magnitude of the historical gain change.

[0097] By calculating the historical gain change rate and the historical feedback layer gain, the performance change of the electricity meter can be quantitatively evaluated.

[0098] Step S27: Determine the final gain parameter according to the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain, and historical feedback layer gain for the current sampling period.

[0099] In a preferred embodiment, determining the final gain parameter according to the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain, and historical feedback layer gain for the current sampling period includes:

[0100] The final gain parameter is calculated according to the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain, and historical feedback layer gain for the current sampling period through the following formula:

[0101]

[0102] G(t) = W d ·G d +W TH ·G TH +W h ·G h ;

[0103] where W d represents the interference layer weight for the current sampling period, S d represents the interference intensity factor for the current sampling period, K T represents the temperature factor for the current sampling period, K H represents the humidity factor for the current sampling period, W TH represents the temperature and humidity layer weight for the current sampling period, W h represents the historical feedback layer weight for the current sampling period, G d represents the interference layer gain for the current sampling period, GTH The temperature and humidity layer gain representing the current sampling period, and G(t) represents the final gain parameter.

[0104] For step S27, an adaptive weighting mechanism is introduced. According to the temperature and humidity layer gain, historical feedback layer gain, interference layer gain, temperature factor, humidity factor, absolute value of the historical gain change rate, and interference intensity factor, the weights of the gains of each layer are dynamically adjusted, and the final gain parameter is calculated based on the sum of the products of the weights of each layer and the corresponding layer gains.

[0105] During the adjustment process of the entire gain parameter, the main frequency interference component of the current window is extracted through fast Fourier transform (FFT), which can accurately identify the main interference frequency in the environment, and combined with the voltage and current fluctuation amplitudes to calculate the interference intensity factor. It can sensitively detect the source of external electromagnetic interference during the operation of the electric energy meter and quantify its intensity. The introduction of the interference layer gain enables the system to adaptively adjust the gain according to the interference intensity factor, ensuring that the gain is more sensitive when the interference is small and tends to be stable when the interference is large, avoiding system instability caused by excessive gain. This can effectively improve the fault detection accuracy of the electric energy meter in different interference environments. Through the introduction of the temperature and humidity layer gain, the system can more accurately cope with the influence of temperature and humidity on the metering accuracy of the electric energy meter, ensuring that the electric energy meter can still maintain the measurement accuracy when the temperature and humidity change greatly, thereby improving the applicability and stability of the system. The historical feedback layer gain can be adaptively adjusted according to the gain change rate, so as to automatically adjust the response speed of the system when the gain fluctuation is detected. This can quickly respond after a fault or interference occurs and reduce the error caused by delay. By considering different influencing factors such as interference, environmental temperature and humidity, and historical feedback, the weight ratio of the gains of each layer can be adaptively adjusted to ensure the coordination and optimization effect of the gains of each layer and avoid the instability caused by gain imbalance.

[0106] Step S3: Multiply the final gain parameter by the current bandwidth and the current cut-off frequency of the anti-interference filter respectively to obtain the updated bandwidth and the updated cut-off frequency of the anti-interference filter;

[0107] For step S3, the anti-interference filter adjusts the bandwidth and cut-off frequency of the anti-interference filter based on the final gain parameter to adapt to different intensities of interference. Specifically, multiply the final gain parameter by the current bandwidth and cut-off frequency respectively as the updated bandwidth and cut-off frequency, and then the filter settings can be dynamically adjusted according to different intensities of interference, so as to filter out interference more effectively. This method of dynamically adjusting the filter settings can enable the filter to maintain effective filtering performance under different intensities of interference. For example, compared with the previous sampling moment, the voltage fluctuation amplitude ΔV at the current sampling moment is larger, that is, when the interference is stronger, the interference intensity factor S dIt increases correspondingly, resulting in a relatively large proportion of the interference layer weight in the current sampling period, and then increasing the final gain parameter G(t). At this time, the increased final gain parameter is multiplied by the current bandwidth and cut-off frequency, thereby increasing the bandwidth and cut-off frequency of the filter. Such an adjustment can make the filter pass the useful signal more tolerantly and filter out the interference signal more effectively. Because the increased bandwidth allows more frequency components to pass through, and the increased cut-off frequency can better suppress high-frequency interference.

[0108] Step S4: When the anti-interference filter operates based on the updated bandwidth and updated cut-off frequency, obtain the voltage fluctuation value and current fluctuation value filtered by the anti-interference filter;

[0109] For step S4, the anti-interference filter filters the input signal according to the actual interference characteristics and the working requirements of the electricity meter, effectively filtering out some non-faulty interferences, avoiding the influence of normal fluctuations and other interference factors on the discrimination accuracy of high-interference faults, thereby reducing false alarms and improving the reliability and stability of the electricity meter in a high-interference environment. By detecting the filtered voltage and current fluctuation values in real time, the actual working state of the electricity meter after anti-interference processing can be obtained, providing an accurate data basis for subsequent fault judgment.

[0110] Step S5: Compare the voltage fluctuation value with the voltage fluctuation threshold, and compare the current fluctuation value with the current fluctuation threshold. If the voltage fluctuation value is greater than the voltage fluctuation threshold and the current fluctuation value is greater than the current fluctuation threshold, it is determined that the electricity meter has a high-interference fault; otherwise, it is determined that the electricity meter does not have a high-interference fault.

[0111] For step S5, by comparing the fluctuation values before and after filtering, the processing effect of the anti-interference filter can be evaluated to ensure that while filtering out the interference, the true high-interference faults can be accurately identified. When the filtered voltage fluctuation value is greater than the voltage fluctuation threshold and the current fluctuation value is greater than the current fluctuation threshold, it indicates that even after filtering, the electricity meter is still affected by high interference. In this case, it can be determined that the electricity meter has a high-interference fault and the fault source needs to be further located.

[0112] It should be noted that the current fluctuation value and the current fluctuation threshold can be determined based on the historical mean and standard deviation, which can more sensitively identify abnormal situations in a high-interference environment.

[0113] In a preferred embodiment, after determining that the electricity meter has a high-interference fault, it further includes:

[0114] Take the electricity meter determined to have a high-interference fault as the target electricity meter;

[0115] Establish a multivariate Gaussian distribution model based on the voltage data of the target electricity meter;

[0116] Based on the multivariate Gaussian distribution model, the log-likelihood value of the voltage data of the target electricity meter is calculated;

[0117] The log-likelihood value of the voltage data of the target electricity meter is compared with a preset voltage threshold;

[0118] If the log-likelihood value of the voltage data of the target electricity meter is greater than or equal to the preset voltage threshold, it is determined that the voltage measurement module in the target electricity meter is faulty;

[0119] If the log-likelihood value of the voltage data of the target electricity meter is less than the preset voltage threshold, the fault type and probability of the target electricity meter are determined through a fault identification model.

[0120] In an embodiment of the present invention, based on the electricity meters determined to have high-interference faults, the following multivariate Gaussian distribution model is established using standardized voltage data:

[0121]

[0122] where f(V′) represents the probability density of the standardized voltage data vector V′, ∑ represents the covariance matrix of the voltage data, d represents the number of electricity meters, μ represents the mean of the standardized voltage data, (V′ - μ)′∑ -1 (V′ - μ) represents the Mahalanobis distance. Different from the Euclidean distance, the Mahalanobis distance takes into account the variance and correlation of the data, so it is more suitable for multivariate distributions;

[0123] Based on the multivariate Gaussian distribution model, according to the real-time collected standardized voltage data, the log-likelihood value is calculated through the following formula:

[0124]

[0125] where, logL(V t ′) represents the log-likelihood value of the standard voltage data V t ′ at time t, log|∑| represents the logarithm of the determinant of the covariance matrix ∑, |∑| represents the determinant of the covariance matrix, which reflects the volume or extent of the covariance matrix. The larger the value, the wider the "range" of the data distribution, and the smaller the value, the more concentrated the data. Through the logarithmic operation of log|∑|, the log-likelihood function can be made smoother and at the same time avoid numerical overflow problems in numerical calculations;

[0126] Based on historical data, a preset voltage threshold for the log-likelihood value is determined and a preliminary quick screening is performed. If the log-likelihood value calculated in real time is greater than or equal to the preset voltage threshold, it is determined that the voltage measurement module in the electricity meter is faulty. If the log-likelihood value is less than the voltage threshold, it is determined that further fault identification is required.

[0127] It should be noted that by establishing a multivariate Gaussian distribution model, the present invention fully considers the variance and correlation of voltage data, and uses the Mahalanobis distance to measure the degree of data deviation from the normal state. Compared with the Euclidean distance that only considers the independence of each variable, the Mahalanobis distance is more suitable for multivariate distributions and can more accurately detect abnormal data points. In the case of high-interference faults, using standardized voltage data to establish a model can effectively reduce the influence of different dimensions and scales, making the model more robust. By using standardized voltage data to establish a multivariate Gaussian distribution model and combining the calculation of the log-likelihood value and threshold comparison, efficient and accurate preliminary screening of energy meters with high-interference faults is achieved, improving the accuracy and efficiency of fault detection, enhancing the system's adaptability to complex interference environments, and laying a solid foundation for further fault identification and maintenance work.

[0128] Preferably, through the fault identification model, determine the fault type and probability of the target energy meter, including:

[0129] Perform weighted processing on the fault characteristics of the target energy meter according to the preset optimal weight combination to obtain the weighted fault characteristics; the fault characteristics include: the voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate, and final gain parameter in the current sampling period;

[0130] Input the weighted fault characteristics into the trained fault type recognition model, so that the fault recognition model generates each fault type of the target energy meter and its corresponding probability according to the weighted fault characteristics.

[0131] Specifically, for energy meters determined to have high-interference faults but with the voltage measurement module judged to be fault-free through the log-likelihood value, it is next necessary to further determine the fault type and probability through the fault recognition model. Specifically, in the fault recognition model, first perform weighted processing on the fault characteristics of the target energy meter according to the preset optimal weight combination. The fault characteristics include multiple aspects, such as the voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate, and final gain parameter in the current sampling period, etc. These characteristics can comprehensively reflect the working state of the energy meter and possible fault conditions. The purpose of weighted processing is to assign different weights to each fault characteristic according to its importance, so as to more accurately highlight those characteristics that have a key impact on the judgment of fault type and probability, thereby improving the accuracy and reliability of fault recognition. Input the weighted fault characteristics into the trained fault type recognition model. The fault type recognition model performs complex calculations and analyses based on the input weighted fault characteristics, generates each fault type of the target energy meter and its corresponding probability, and provides a clear basis for fault judgment.

[0132] Next, the weighted fault features are input into the trained fault type recognition model. This fault type recognition model includes an input layer, a fuzzy membership layer, a rule layer, a normalization layer, and an output layer. The input layer inputs the fault features, including the voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate, and final gain parameter in the current sampling period. The fuzzy membership layer uses triangular membership functions to assign the input features to multiple membership sets and calculates the fuzzy membership values through the membership functions. The rule layer applies fuzzy rules based on the input fuzzy membership values. The normalization layer normalizes the output of the rule layer. The output layer uses the Takagi-Sugeno model to output the final fault probability through the output layer by combining the normalized rule output with a linear equation.

[0133] To obtain this accurate fault type recognition model, an optimal weight combination is required.

[0134] Preferably, the preset optimal weight combination is determined in the following manner:

[0135] Obtain a number of training samples; among them, each training sample includes: the historical fault features of the electric energy meter and its corresponding actual faults; the historical fault features include: the voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate, and final gain parameter in the historical sampling period.

[0136] Determine the number of weight combinations according to the preset population size.

[0137] Randomly initialize each weight combination to obtain a number of initialized weight combinations; among them, the weight combination includes the weight value of each feature in the historical fault features.

[0138] Repeat the wild horse optimization process until the preset maximum number of iterations is reached, or the loss value of the leading combination converges, to obtain the optimal weight combination.

[0139] The wild horse optimization process includes:

[0140] For each current weight combination, perform weighted processing on the current weight combination and the historical fault features in each training sample to obtain the weighted historical fault features of each training sample; among them, when the wild horse optimization process is executed for the first time, the current weight combination is the initialized weight combination.

[0141] Input the weighted historical fault features of each training sample into the fault type recognition model to be trained, so that the fault type recognition model to be trained takes the weighted historical fault features of each training sample as input and outputs the predicted fault type of each training sample and its corresponding probability. During the training process, calculate the cross-entropy loss function according to the probability of the predicted fault type of each training sample and the actual fault of the corresponding training sample, and adjust the network parameters of the fault type recognition model according to the cross-entropy loss function until the cross-entropy loss function converges to obtain a pre-trained fault type recognition model;

[0142] Take the convergence value of the cross-entropy loss function as the loss value;

[0143] Take the current weight combination with the smallest loss value among all current weight combinations as the leading combination;

[0144] For each current weight combination, update the current weight combination according to the leading combination to obtain the updated weight combination;

[0145] Judge whether the preset maximum number of iterations is satisfied during this wild horse optimization process, or whether the loss value of the leading combination converges. If so, take the leading combination as the optimal weight combination; otherwise, take each updated weight combination as each current weight combination for the next execution of the wild horse optimization process.

[0146] Specifically, the present invention introduces a wild horse optimization algorithm to confirm the optimal weight parameters and train the fault type recognition model, which not only improves the accuracy of fault recognition but also enhances the generalization ability of the model. Specifically, first, determine the population size and the maximum number of iterations of the wild horse optimization algorithm based on historical data. Then, according to the number of items of the fault features, the number of weights can be determined, that is, each input feature item corresponds to a weight value, and these weight values together constitute a weight combination.

[0147] During the wild horse optimization process, define the individuals of the population size, and each individual contains a set of weight combinations. To evaluate the quality of these weight combinations, select the cross-entropy loss function to calculate the loss between the predicted fault probability weighted by the weight combination corresponding to the individual in the population size and the actual label. Through this loss, the impact of each weight combination on the performance of the fault type recognition model can be quantitatively evaluated.

[0148] Next, group the population individuals, weight the fault features according to the weight combination of each individual, and then input these weighted fault features into the fault type recognition model to be trained for model training. During the training process, we continuously adjust the network parameters of the model according to the cross-entropy loss function until the loss function converges, thereby obtaining a pre-trained fault type recognition model.

[0149] In each round of wild horse optimization iteration, the convergence value of the pre-trained fault type recognition model obtained by training each individual is used as the loss value, and the individual with the smallest calculated loss value is selected as the optimal weight combination, that is, the leader combination. Then, according to the weight combination of the leader individual, the weight combinations of other individuals are updated through the following formula in the hope of obtaining a better weight combination in the next round of iteration:

[0150]

[0151] where, represents an updated weight combination, represents a weight combination, r represents a random number, and its value range is [-1, 1], represents the leader combination.

[0152] When the maximum number of iterations of the wild horse optimization algorithm is reached and the loss value no longer decreases significantly, the iteration stops. At this time, the optimized leader combination is used as the optimal weight combination, which represents the weight distribution of each feature item in the fault characteristics that is most critical for fault type recognition. At the same time, based on this optimal weight combination, the pre-trained fault type recognition model obtained during this process has been verified by a large number of training samples and can accurately map the weighted fault characteristics to the corresponding fault types and probabilities. Therefore, this pre-trained model trained based on the leader combination is used as the trained fault type recognition model.

[0153] By introducing the wild horse optimization algorithm, the present invention realizes the automatic search and determination of the optimal weight parameters, avoids misjudgment of the model when specific features have a greater impact, and further enhances the fault prediction accuracy. Through weight optimization, the most influential feature items can be identified, which is convenient for accurately identifying the key factors most influential on faults, helps subsequent fault location and maintenance decision-making. It can not only accurately diagnose the electricity meters with further identified faults, but also can adaptively respond to various interference factors through adaptive feature weights and adaptive threshold settings.

[0154] In addition, through a multi-level screening method, that is, preliminary screening after gain adjustment, logarithmic likelihood value screening, and in-depth analysis of the fault type recognition model, normal signals or minor interferences are gradually screened and eliminated to avoid the occupation of system resources by irrelevant abnormalities, and it can effectively reduce false alarms and missed alarms caused by environmental fluctuations or instantaneous interferences.

[0155] Preferably, the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain, historical feedback layer gain, final gain parameter, high-interference fault discrimination result, voltage measurement module fault discrimination result, and prediction result of the fault recognition model in the current sampling period are sent to the visualization device for visualization display by the visualization device.

[0156] Preferably, when there is a high-interference fault, a voltage measurement module fault, or the fault probability generated by the fault recognition model exceeds the alarm threshold, an alarm message is sent through a wireless device, and an alarm is issued through an audible and visual alarm.

[0157] As Figure 3 shown, based on the above method item embodiments, corresponding device item embodiments are provided;

[0158] An embodiment of the present invention provides an electric energy meter fault detection device, including: a sampling data acquisition module, a gain parameter calculation module, a filter parameter update module, a fluctuation value acquisition module, and a high-interference fault discrimination module;

[0159] The sampling data acquisition module is used to acquire voltage data, current data, and temperature and humidity data of the electric energy meter in the current sampling period, and the historical feedback layer gain of the previous sampling period;

[0160] The gain parameter calculation module is used to determine the final gain parameter according to the voltage data, current data, and temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period;

[0161] The filter parameter update module is used to multiply the final gain parameter by the current bandwidth and the current cut-off frequency of the anti-interference filter respectively to obtain the updated bandwidth and the updated cut-off frequency of the anti-interference filter;

[0162] The fluctuation value acquisition module is used to acquire the voltage fluctuation value and the current fluctuation value filtered by the anti-interference filter when the anti-interference filter works based on the updated bandwidth and the updated cut-off frequency;

[0163] The high-interference fault discrimination module is used to compare the voltage fluctuation value with the voltage fluctuation threshold, and compare the current fluctuation value with the current fluctuation threshold. If the voltage fluctuation value is greater than the voltage fluctuation threshold and the current fluctuation value is greater than the current fluctuation threshold, it is determined that the electric energy meter has a high-interference fault. Otherwise, it is determined that the electric energy meter does not have a high-interference fault.

[0164] It can be understood that the above device item embodiments correspond to the method item embodiments of the present invention, and can implement the electric energy meter fault detection method provided by any one of the above method item embodiments of the present invention.

[0165] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationships between the modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement this without creative effort.

[0166] Based on the embodiments of the above power meter fault detection method, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the power meter fault detection method of any embodiment of the present invention is implemented.

[0167] Exemplarily, in this embodiment, the computer program can be divided into one or more modules, and the one or more modules are stored in the memory and executed by the processor to complete the present invention. The one or more module elements can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program in the terminal device.

[0168] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0169] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device, and connects various parts of the entire terminal device through various interfaces and lines.

[0170] Based on the above method embodiment, another embodiment is provided: A computer-readable storage medium provided by another embodiment of the present invention includes a stored computer program, wherein when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the electricity meter fault detection method described in any one of the above method embodiments of the present invention.

[0171] Among them, if the module / unit integrated in the electricity meter fault detection device / terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0172] The above is the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A method for detecting faults in an electric energy meter, characterized in that: include: Obtain the voltage data, current data, temperature and humidity data of the electric energy meter during the current sampling period, as well as the historical feedback layer gain of the previous sampling period; Determine the final gain parameter based on the voltage data, current data, temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period; Multiplying the final gain parameter by the current bandwidth and the current cutoff frequency of the anti-interference filter respectively to obtain an updated bandwidth and an updated cutoff frequency of the anti-interference filter; When the anti-interference filter works based on the updated bandwidth and the updated cut-off frequency, the voltage fluctuation value and the current fluctuation value after being filtered by the anti-interference filter are obtained; The voltage fluctuation value is compared with the voltage fluctuation threshold, and the current fluctuation value is compared with the current fluctuation threshold. If the voltage fluctuation value is greater than the voltage fluctuation threshold, and the current fluctuation value is greater than the current fluctuation threshold, it is judged that the electric energy meter has a high interference fault; otherwise, it is judged that the electric energy meter does not have a high interference fault.

2. The electric energy meter fault detection method according to claim 1, characterized in that: According to the voltage data, current data, temperature and humidity data of the current sampling period, and the historical feedback layer gain of the previous sampling period, the final gain parameters are determined, including: According to the voltage data, current data and temperature and humidity data of the current sampling period, the voltage fluctuation amplitude, current fluctuation amplitude, temperature factor and humidity factor are calculated; Performing fast Fourier transform on the voltage data, current data and temperature and humidity data, and selecting the frequency component with the highest spectrum density as the environmental interference frequency; Dividing the sum of the voltage fluctuation amplitude and the current fluctuation amplitude by the environmental interference frequency to obtain an interference intensity factor; Determining an interference layer gain according to the interference intensity factor; The temperature factor and humidity factor are summed and averaged to obtain the temperature and humidity layer gain; According to the historical feedback layer gain of the previous sampling period, the historical gain change rate and the historical feedback layer gain of the current sampling period are calculated; The final gain parameter is determined according to the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain and historical feedback layer gain of the current sampling period.

3. The electric energy meter fault detection method according to claim 2, characterized in that: According to the historical feedback layer gain of the previous sampling period, the historical gain change rate and the historical feedback layer gain of the current sampling period are calculated, including: According to the historical feedback layer gain of the previous sampling period, the historical gain change rate and historical feedback layer gain of the current sampling period are calculated by the following formula: G h =1+|R G |; Among them, R G Indicates the historical gain change rate during the current sampling period, G ′ h (tT w ) represents the historical feedback layer gain of the previous sampling period, G ′ h (t-2·T w ) represents the historical feedback layer gain in the previous two sampling periods, t represents the current sampling time, T w Represents the sampling window length, G h Represents the historical feedback layer gain of the current sampling period.

4. The electric energy meter fault detection method according to claim 3, characterized in that: According to the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain and historical feedback layer gain of the current sampling period, the final gain parameters are determined, including: According to the interference intensity factor, temperature factor, humidity factor, historical gain change rate, interference layer gain, temperature and humidity layer gain and historical feedback layer gain of the current sampling period, the final gain parameter is calculated by the following formula: G(t)=W d ·G d +W TH ·G TH +W h ·G h ; Among them, W d represents the interference layer weight of the current sampling period, S d Indicates the interference intensity factor of the current sampling period, K T Indicates the temperature factor of the current sampling period, K H Indicates the humidity factor of the current sampling period, W TH Represents the temperature and humidity layer weight of the current sampling period, W h represents the historical feedback layer weight of the current sampling period, G d represents the interference layer gain of the current sampling period, G TH represents the temperature and humidity layer gain of the current sampling period, and G(t) represents the final gain parameter.

5. The electric energy meter fault detection method according to claim 2, characterized in that: After judging whether the electric energy meter has a high interference fault, it also includes: The electric energy meter judged to be a high interference fault is used as the target electric energy meter; A multivariate Gaussian distribution model is established according to the voltage data of the target electric energy meter; Based on the multivariate Gaussian distribution model, a log-likelihood value of the target electric energy meter voltage data is calculated; Comparing the log-likelihood value of the target electric energy meter voltage data with a preset voltage threshold; If the log-likelihood value of the target electric energy meter voltage data is greater than or equal to the preset voltage threshold, it is determined that the voltage measurement module in the target electric energy meter is faulty; If the log-likelihood value of the target electric energy meter voltage data is less than the preset voltage threshold, the fault type and probability of the target electric energy meter are determined through the fault identification model.

6. The electric energy meter fault detection method according to claim 5, characterized in that: The fault identification model is used to determine the fault type and probability of the target electric energy meter, including: The fault characteristics of the target electric energy meter are weighted according to a preset optimal weight combination to obtain weighted fault characteristics; the fault characteristics include: voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate and final gain parameter in the current sampling period; The weighted fault features are input into a trained fault type recognition model, so that the fault recognition model generates various fault types of the target electric energy meter and their corresponding probabilities according to the weighted fault features.

7. The electric energy meter fault detection method according to claim 6, characterized in that: The preset optimal weight combination is determined in the following manner: Acquire a number of training samples; wherein each training sample includes: historical fault characteristics of the electric energy meter and its corresponding actual fault; the historical fault characteristics include: voltage fluctuation amplitude, current fluctuation amplitude, temperature factor, humidity factor, historical gain change rate and final gain parameter during the historical sampling period; Determine the number of weight combinations based on the preset population size; Each weight combination is randomly initialized to obtain a number of initialized weight combinations; wherein the weight combination includes a weight value of each feature in the historical fault features; Repeat the wild horse optimization process until the preset maximum number of iterations is reached or the loss value of the leader combination converges, and the optimal weight combination is obtained; The Mustang optimization process includes: For each current weight combination, the current weight combination and the historical fault features in each training sample are weighted to obtain the weighted historical fault features of each training sample; wherein, when the Mustang optimization process is executed for the first time, the current weight combination is the initialized weight combination; The weighted historical fault features of each training sample are input into the fault type recognition model to be trained, so that the fault type recognition model to be trained is trained with the weighted historical fault features of each training sample as input and the predicted fault type of each training sample and its corresponding probability as output, and in the training process, the cross entropy loss function is calculated according to the probability of the predicted fault type of each training sample and the actual fault of the corresponding training sample, and the network parameters of the fault type recognition model are adjusted according to the cross entropy loss function until the cross entropy loss function converges, thereby obtaining a pre-trained fault type recognition model; The convergence value of the cross entropy loss function is used as the loss value; The current weight combination with the smallest loss value among all current weight combinations is taken as the leading combination; For each current weight combination, the current weight combination is updated according to the leader combination to obtain an updated weight combination; Determine whether the preset maximum number of iterations is met during this Mustang optimization process, or whether the loss value of the leading combination converges. If so, the leading combination is used as the optimal weight combination; otherwise, each updated weight combination is used as each current weight combination for the next execution of the Mustang optimization process.

8. An electric energy meter fault detection device, characterized in that: include: Sampling data acquisition module, gain parameter calculation module, filter parameter update module, fluctuation value acquisition module and high interference fault identification module; The sampling data acquisition module is used to obtain the voltage data, current data and temperature and humidity data of the electric energy meter in the current sampling period, and the historical feedback layer gain in the previous sampling period; The gain parameter calculation module is used to determine the final gain parameter according to the voltage data, current data and temperature and humidity data of the current sampling period and the historical feedback layer gain of the previous sampling period; The filtering parameter updating module is used to multiply the final gain parameter with the current bandwidth and the current cutoff frequency of the anti-interference filter respectively to obtain an updated bandwidth and an updated cutoff frequency of the anti-interference filter; The fluctuation value acquisition module is used to obtain the voltage fluctuation value and the current fluctuation value after being filtered by the anti-interference filter when the anti-interference filter works based on the updated bandwidth and the updated cut-off frequency; The high-interference fault judgment module is used to compare the voltage fluctuation value with the voltage fluctuation threshold, and to compare the current fluctuation value with the current fluctuation threshold. If the voltage fluctuation value is greater than the voltage fluctuation threshold, and the current fluctuation value is greater than the current fluctuation threshold, it is judged that the electric energy meter has a high-interference fault; otherwise, it is judged that the electric energy meter does not have a high-interference fault.

9. A terminal device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the electric energy meter fault detection method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: include: A stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the electric energy meter fault detection method according to any one of claims 1 to 7.

Images