Electric energy meter fault diagnosis method and device based on multi-dimensional characteristics, equipment and medium

Abstract

The application relates to an electric energy meter fault diagnosis method, device, equipment and medium based on multi-dimensional characteristics. The method comprises the following steps: acquiring electric energy meter characteristics of a to-be-detected electric energy meter in at least one dimension; performing feature extraction on the electric energy meter characteristics in each dimension respectively to obtain characteristic features of the to-be-detected electric energy meter in the corresponding dimension; performing fault detection on the to-be-detected electric energy meter according to the characteristic features of the to-be-detected electric energy meter in at least one dimension through a pre-trained fault detection model to obtain initial fault probabilities of the to-be-detected electric energy meter in at least one fault category; the fault detection model is trained based on a multi-layer perception neural network; and the fault diagnosis result of the to-be-detected electric energy meter is determined according to the initial fault probabilities in at least one fault category, historical probability distribution and category weights of the corresponding fault categories. The method can realize efficient and accurate diagnosis of electric energy meter faults.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault detection, in particular to a power meter fault diagnosis method, device, equipment and medium based on multi-dimensional characteristics. BACKGROUND

[0002] With the continuous development of smart grids, the application of power meters is showing a rapid growth trend. The number of faulty and returned power meters is increasing rapidly every year, which is causing increasing pressure on asset management, and batch analysis of faulty and returned power meters is facing severe challenges.

[0003] The current power meter fault analysis system is still dominated by phenomenon judgment. Related standards focus more on identifying explicit faults such as error and stop, but there is a significant gap in tracing the internal causes of faults. Although there has been some accumulation in the fault evaluation of power meter components, such as fault data collection, fault mode identification and failure analysis, the causal relationship mechanism between fault phenomena and component failure has not been systematically established, making it difficult to achieve reverse reasoning of fault causality. For returned power meters, although existing automatic sorting technology can initially distinguish between tables to be tested and non-recyclable tables, it cannot handle complex fault scenarios and still relies on manual opening detection.

[0004] Although this way can detect the fault of the power meter, this way is subject to the experience difference of the technical personnel, and the detection efficiency is low and the result consistency is poor, thereby causing the problem of inaccurate detection of the power meter. SUMMARY

[0005] Therefore, it is necessary to provide a power meter fault diagnosis method, device, equipment and medium based on multi-dimensional characteristics, which can accurately detect the fault of the power meter.

[0006] In a first aspect, the present application provides a power meter fault diagnosis method based on multi-dimensional characteristics, comprising:

[0007] Obtaining power meter characteristics of a to-be-detected power meter in at least one dimension; the power meter characteristics include impedance characteristics, power characteristics, shape characteristics, regular performance characteristics and running state characteristics;

[0008] Respectively extracting features of the power meter characteristics in each dimension to obtain characteristic features of the to-be-detected power meter in the corresponding dimension;

[0009] According to the characteristic features of the to-be-detected power meter in at least one dimension, performing fault detection on the to-be-detected power meter through a pre-trained fault detection model to obtain an initial fault probability of the to-be-detected power meter in at least one fault category; the fault detection model is trained based on a multi-layer perception neural network;

[0010] According to the initial fault probability under at least one fault category, the historical probability distribution, and the category weight of the corresponding fault category, a fault diagnosis result of the to-be-detected electric energy meter is determined.

[0011] In one of the embodiments, according to the characteristic features of the to-be-detected electric energy meter in at least one dimension, the to-be-detected electric energy meter is subjected to fault detection by a pre-trained fault detection model to obtain an initial fault probability of the to-be-detected electric energy meter under at least one fault category, which includes:

[0012] According to the characteristic features of the to-be-detected electric energy meter in at least one dimension, the characteristic features are subjected to feature enhancement processing to obtain corresponding feature enhancement processed characteristic features.

[0013] For each fault category, according to the feature enhancement processed characteristic features, a preset balance parameter, and the total number of fault categories, the initial fault probability of the to-be-detected electric energy meter under the fault category is determined.

[0014] In one of the embodiments, according to the characteristic features of the to-be-detected electric energy meter in at least one dimension, the characteristic features are subjected to feature enhancement processing to obtain corresponding feature enhancement processed characteristic features, which includes:

[0015] According to the characteristic features of the to-be-detected electric energy meter in at least one dimension, an average feature of the to-be-detected electric energy meter is determined.

[0016] For the characteristic features in each dimension, according to the characteristic features and the corresponding average features, the characteristic features are subjected to feature enhancement processing to obtain corresponding feature enhancement processed characteristic features.

[0017] In one of the embodiments, according to the feature enhancement processed characteristic features, a preset balance parameter, and the total number of fault categories, the initial fault probability of the to-be-detected electric energy meter under the fault category is determined, which includes:

[0018] Based on an output layer in the fault detection model, according to the feature enhancement processed characteristic features and a preset balance parameter, an initial fault value of the to-be-detected electric energy meter under the fault category is determined; and,

[0019] According to the feature enhancement processed characteristic features, a preset balance parameter, and the total number of fault categories, a total fault value of the to-be-detected electric energy meter is determined.

[0020] A ratio between the initial fault value and the total fault value is taken as the initial fault probability of the to-be-detected electric energy meter under the fault category.

[0021] In one of the embodiments, according to the initial fault probability under at least one fault category, the historical probability distribution, and the category weight of the corresponding fault category, a fault diagnosis result of the to-be-detected electric energy meter is determined, which includes:

[0022] For each fault category, according to the confidence of the fault category to the electric energy meter to be detected, a category weight of the fault category is determined;

[0023] The initial fault probability under each fault category, the historical probability distribution and the category weight of the corresponding fault category are weighted to obtain a fault diagnosis result of the electric energy meter to be detected.

[0024] In one of the embodiments, the method further comprises:

[0025] The electric energy meter appearance, the routine performance detection data, the running state word, the impedance Bode diagram, the electric quantity gray scale diagram and the fault diagnosis result of the electric energy meter to be detected are displayed in the visual interface;

[0026] The impedance Bode diagram is converted based on the impedance characteristics of the electric energy meter to be detected; the electric quantity gray scale diagram is converted based on the electric quantity characteristics of the electric energy meter to be detected; the electric energy meter appearance is converted based on the appearance characteristics of the electric energy meter to be detected; the routine performance detection data is converted based on the routine performance characteristics of the electric energy meter to be detected; and the running state word is converted based on the running state characteristics of the electric energy meter to be detected.

[0027] In a second aspect, the present application further provides an electric energy meter fault diagnosis device, comprising:

[0028] The acquisition module is configured to acquire electric energy meter characteristics of the electric energy meter to be detected in at least one dimension; the electric energy meter characteristics include impedance characteristics, electric quantity characteristics, appearance characteristics, routine performance characteristics and running state characteristics;

[0029] The enhancement module is configured to perform feature extraction on the electric energy meter characteristics in each dimension respectively to obtain characteristic features of the electric energy meter to be detected in the corresponding dimension;

[0030] The recognition module is configured to perform fault detection on the electric energy meter to be detected by using a pre-trained fault detection model according to the characteristic features of the electric energy meter to be detected in at least one dimension to obtain an initial fault probability of the electric energy meter to be detected in at least one fault category; the fault detection model is trained based on a multilayer perception neural network;

[0031] The diagnosis module is configured to determine a fault diagnosis result of the electric energy meter to be detected according to the initial fault probability in at least one fault category, the historical probability distribution and the category weight of the corresponding fault category.

[0032] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0033] acquire the electric energy meter characteristics of the to-be-detected electric energy meter in at least one dimension; the electric energy meter characteristics include impedance characteristics, electric quantity characteristics, shape characteristics, general performance characteristics, and running state characteristics;

[0034] perform feature extraction on the electric energy meter characteristics in each dimension respectively to obtain the characteristic features of the to-be-detected electric energy meter in the corresponding dimension;

[0035] perform fault detection on the to-be-detected electric energy meter according to the characteristic features of the to-be-detected electric energy meter in at least one dimension through a pre-trained fault detection model to obtain initial fault probabilities of the to-be-detected electric energy meter in at least one fault category; the fault detection model is trained based on a multilayer perceptron neural network;

[0036] determine a fault diagnosis result of the to-be-detected electric energy meter according to the initial fault probabilities in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category.

[0037] In a fourth aspect, the present application further provides a computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the following steps:

[0038] acquire the electric energy meter characteristics of the to-be-detected electric energy meter in at least one dimension; the electric energy meter characteristics include impedance characteristics, electric quantity characteristics, shape characteristics, general performance characteristics, and running state characteristics;

[0039] perform feature extraction on the electric energy meter characteristics in each dimension respectively to obtain the characteristic features of the to-be-detected electric energy meter in the corresponding dimension;

[0040] perform fault detection on the to-be-detected electric energy meter according to the characteristic features of the to-be-detected electric energy meter in at least one dimension through a pre-trained fault detection model to obtain initial fault probabilities of the to-be-detected electric energy meter in at least one fault category; the fault detection model is trained based on a multilayer perceptron neural network;

[0041] determine a fault diagnosis result of the to-be-detected electric energy meter according to the initial fault probabilities in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category.

[0042] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which is executed by a processor to implement the following steps:

[0043] acquire the electric energy meter characteristics of the to-be-detected electric energy meter in at least one dimension; the electric energy meter characteristics include impedance characteristics, electric quantity characteristics, shape characteristics, general performance characteristics, and running state characteristics;

[0044] perform feature extraction on the electric energy meter characteristics in each dimension respectively to obtain the characteristic features of the to-be-detected electric energy meter in the corresponding dimension;

[0045] According to the characteristic features of the to-be-detected electric energy meter in at least one dimension, the to-be-detected electric energy meter is subjected to fault detection through a pre-trained fault detection model, to obtain an initial fault probability of the to-be-detected electric energy meter in at least one fault category; the fault detection model is trained based on a multilayer perceptron neural network;

[0046] According to the initial fault probability in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category, a fault diagnosis result of the to-be-detected electric energy meter is determined.

[0047] The above electric energy meter fault diagnosis method, device, equipment and medium based on multi-dimensional characteristics obtain electric energy meter characteristics of the to-be-detected electric energy meter in at least one dimension; the electric energy meter characteristics include impedance characteristics, electric quantity characteristics, shape characteristics, conventional performance characteristics, and running state characteristics; the electric energy meter characteristics in each dimension are subjected to feature extraction respectively, to obtain characteristic features of the to-be-detected electric energy meter in the corresponding dimension; according to the characteristic features of the to-be-detected electric energy meter in at least one dimension, the to-be-detected electric energy meter is subjected to fault detection through a pre-trained fault detection model, to obtain an initial fault probability of the to-be-detected electric energy meter in at least one fault category; the fault detection model is trained based on a multilayer perceptron neural network; according to the initial fault probability in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category, a fault diagnosis result of the to-be-detected electric energy meter is determined. This embodiment solves the problems of the existing electric energy meter fault diagnosis, such as dependence on a single data source, insufficient real-time performance, high dependence on artificial experience, and weak correlation analysis capability for complex faults, by means of multi-source information fusion technology. By integrating electric energy meter characteristics, the accuracy of electric energy meter fault diagnosis is significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating any inventive labor.

[0049] Figure 1 An application environment diagram of an electric energy meter fault diagnosis method provided by the present embodiment;

[0050] Figure 2 A flowchart of a first electric energy meter fault diagnosis method provided by the present embodiment;

[0051] Figure 3 A flowchart of a feature enhancement step provided by the present embodiment;

[0052] Figure 4 A flowchart of an initial fault probability determination step provided for the embodiment is shown in the figure;

[0053] Figure 5A A flowchart of a fault diagnosis result determination step provided for the embodiment is shown in the figure;

[0054] Figure 5B A fault diagnosis result diagram provided for the embodiment is shown in the figure;

[0055] Figure 6 A structural block diagram of an electric energy meter fault diagnosis apparatus provided for the embodiment is shown in the figure;

[0056] Figure 7 An internal structure diagram of a computer device provided for the embodiment is shown in the figure. DETAILED DESCRIPTION

[0057] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0058] The electric energy meter fault diagnosis method based on multi-dimensional characteristics provided by the embodiments of the present application can be applied in an application environment as shown in the figure. Figure 1 The computer device 101 obtains electric energy meter characteristics of the to-be-detected electric energy meter 102 in at least one dimension; feature extraction is performed on the electric energy meter characteristics in each dimension respectively to obtain characteristic features of the to-be-detected electric energy meter in the corresponding dimension; fault detection is performed on the to-be-detected electric energy meter by a pre-trained fault detection model according to the characteristic features of the to-be-detected electric energy meter in at least one dimension to obtain initial fault probabilities of the to-be-detected electric energy meter in at least one fault category; the fault detection model is trained based on a multi-layer perception neural network; and a fault diagnosis result of the to-be-detected electric energy meter is determined according to the initial fault probabilities in at least one fault category and category weights of the corresponding fault categories. The computer device can be a terminal or a server. The terminal can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The Internet of Things device can be a smart speaker, a smart television, a smart air conditioner, a smart vehicle-mounted device, etc. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0059] In an exemplary embodiment, as shown in the figure, Figure 2 an electric energy meter fault diagnosis method based on multi-dimensional characteristics is provided. The method is applied in Figure 1The computer device 101 in the embodiment is taken as an example for description, including the following steps S201 to S204. Wherein:

[0060] S201 obtaining the electric energy meter characteristics of the to-be-detected electric energy meter in at least one dimension.

[0061] The electric energy meter characteristics can include at least one of impedance characteristics, electric quantity characteristics, shape characteristics, general performance characteristics and running state characteristics.

[0062] In some embodiments, the general performance characteristics of the to-be-detected electric energy meter are obtained by appearance inspection, alternating current voltage test, hidden movement test, starting test, basic error test, clock test, etc.; the impedance characteristics, electric quantity characteristics and running state of the to-be-detected electric energy meter are obtained by impedance test, harmonic injection test and communication test.

[0063] It should be noted that the embodiment can also be provided with a basic information library, a disassembly information library and an expert experience library to assist the computer device in detecting the fault of the to-be-detected electric energy meter. The basic information library can be constructed by importing fields such as asset number, model, device type, manufacturer information, installation time, etc. from external forms; the disassembly information library can include the correspondence between fault phenomena and fault components; the expert experience library can include a large amount of historical data of fault phenomena-fault components, and the data in the expert experience library can be imported from the disassembly information library.

[0064] S202 performing feature extraction on the electric energy meter characteristics in each dimension respectively to obtain the characteristic features of the to-be-detected electric energy meter in the corresponding dimension.

[0065] In some embodiments, the feature extraction is performed on the electric energy meter characteristics in each dimension respectively by various ways such as encoding, pre-classification, etc. to obtain the characteristic features of the to-be-detected electric energy meter in the corresponding dimension.

[0066] S203 performing fault detection on the to-be-detected electric energy meter according to the characteristic features of the to-be-detected electric energy meter in at least one dimension by the pre-trained fault detection model to obtain the initial fault probability of the to-be-detected electric energy meter in at least one fault category.

[0067] The fault detection model is trained based on a multi-layer perception neural network.

[0068] In some embodiments, the characteristic features are processed by feature enhancement according to the characteristic features of the to-be-detected electric energy meter in at least one dimension to obtain the characteristic features after feature enhancement; for each fault category, the initial fault probability of the to-be-detected electric energy meter in the fault category is determined based on the fault detection model according to the characteristic features after feature enhancement, preset balance parameters and the total number of fault categories.

[0069] Specifically, according to the characteristic features of the to-be-detected electric energy meter in at least one dimension, a principal component direction is determined by a principal components analysis (PCA) method; by solving the eigenvalues and eigenvectors of a covariance matrix, a larger characteristic feature is selected as the principal component direction, and each characteristic feature is subjected to feature enhancement processing to obtain a corresponding characteristic feature after feature enhancement processing. For each fault category, the characteristic features after feature enhancement processing, a preset balance parameter, and the total number of fault categories are input into a pre-trained fault detection model to obtain an initial fault probability of the to-be-detected electric energy meter under the fault category.

[0070] It should be noted that the fault detection model can use a multi-layer perceptron neural network for fault diagnosis, and the network structure includes an input layer, a plurality of hidden layers, and an output layer, and the activation function uses ReLU and Softmax. The input layer is obtained by data preprocessing, the hidden layer includes a plurality of neurons, the number of neurons in the output layer is consistent with the number of fault categories, and the Softmax function is used for probabilistic output.

[0071] S204 determines a fault diagnosis result of the to-be-detected electric energy meter according to the initial fault probability under at least one fault category, the historical probability distribution, and the category weight of the corresponding fault category.

[0072] In some embodiments, a weighted sum value between the initial fault probability under at least one fault category, the historical probability distribution, and the category weight of the corresponding fault category is taken as the fault diagnosis result of the to-be-detected electric energy meter.

[0073] It should be noted that the electric energy meter appearance, the conventional performance detection data, the running state word, the impedance Bode diagram, the electric quantity gray scale diagram, and the fault diagnosis result of the to-be-detected electric energy meter are displayed in the visualization interface; wherein the impedance Bode diagram is converted based on the impedance characteristics of the to-be-detected electric energy meter; the electric quantity gray scale diagram is converted based on the electric quantity characteristics of the to-be-detected electric energy meter; the electric energy meter appearance is converted based on the appearance characteristics of the to-be-detected electric energy meter; the conventional performance detection data is converted based on the conventional performance characteristics of the to-be-detected electric energy meter; and the running state word is converted based on the running state characteristics of the to-be-detected electric energy meter. The visualization interface reproduces the appearance of the electric energy meter through a 360° panoramic picture.

[0074] It should be noted that the present embodiment can also identify the electric energy meter appearance of the to-be-detected electric energy meter to extract the model information and the liquid crystal screen state, use the communication state word to extract the state information, use the adaptive hybrid filtering algorithm to denoise the impedance and electric quantity signals, and convert them into the electric quantity gray scale diagram and the impedance Bode diagram.

[0075] The above power meter fault diagnosis method based on multi-dimensional characteristics obtains power meter characteristics of the power meter to be detected in at least one dimension; the power meter characteristics include impedance characteristics, power characteristics, shape characteristics, conventional performance characteristics, and running state characteristics; feature extraction is performed on the power meter characteristics in each dimension respectively to obtain characteristic features of the power meter to be detected in the corresponding dimension; the power meter to be detected is subjected to fault detection through a pre-trained fault detection model according to the characteristic features of the power meter to be detected in at least one dimension, to obtain initial fault probabilities of the power meter to be detected in at least one fault category; the fault detection model is trained based on a multi-layer perception neural network; and a fault diagnosis result of the power meter to be detected is determined according to the initial fault probabilities in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category. Through multi-source information fusion technology, the embodiment solves the problems of the existing power meter fault diagnosis, such as dependence on a single data source, insufficient real-time performance, high dependence on artificial experience, and weak complex fault correlation analysis capability. By integrating power meter characteristics, the accuracy of power meter fault diagnosis is significantly improved.

[0076] Figure 3 For a flowchart of the feature enhancement step in one embodiment, the steps include the following steps:

[0077] S301 determines the average features of the power meter to be detected according to the characteristic features of the power meter to be detected in at least one dimension.

[0078] In some embodiments, the characteristic features of the power meter to be detected in at least one dimension are subjected to mean value processing, and the mean value result is taken as the average features of the power meter to be detected.

[0079] S302, for the characteristic features in each dimension, performs feature enhancement processing on the characteristic features according to the characteristic features and the corresponding average features, to obtain the characteristic features after the feature enhancement processing.

[0080] In some embodiments, for the characteristic features in each dimension, the feature enhancement processing is performed on the characteristic features according to the characteristic features and the corresponding average features through the following formula (1-1), to obtain the characteristic features after the feature enhancement processing.

[0081] (1-1)

[0082] In the formula, represents the characteristic features of the kth power meter to be detected in the ith dimension, represents the average features of the power meter to be detected in the ith dimension, m is the number of power meters to be detected, i represents the ith dimension, and j represents the jth dimension.

[0083] ​In the embodiment, according to the characteristic features of the to-be-detected electric energy meter in at least one dimension, the average features of the to-be-detected electric energy meter are determined; for the characteristic features in each dimension, the characteristic features are subjected to feature enhancement processing according to the characteristic features and the corresponding average features, to obtain the corresponding characteristic features after feature enhancement processing. In the embodiment, the characteristic features after enhancement processing can facilitate accurate processing of the subsequent fault detection model, so as to further improve the accuracy of the output result of the fault detection model.

[0084] Figure 4 For a flowchart of the initial fault probability determination step in one embodiment, the steps include the following:

[0085] S401 determines the initial fault value of the to-be-detected electric energy meter in the fault category based on the output layer in the fault detection model and according to the characteristic features after feature enhancement processing and the preset balance parameter.

[0086] In some embodiments, the ratio between the characteristic features after feature enhancement processing and the preset balance parameter is taken as the initial fault value of the to-be-detected electric energy meter in the fault category.

[0087] S402 determines the total fault value of the to-be-detected electric energy meter according to the characteristic features after feature enhancement processing, the preset balance parameter, and the total number of fault categories.

[0088] In some embodiments, the initial fault values in each fault category are summed to obtain the total fault value of the to-be-detected electric energy meter.

[0089] It should be noted that the step of S401 determining the initial fault value of the to-be-detected electric energy meter in the fault category based on the characteristic features after feature enhancement processing and the preset balance parameter, and the step of S402 determining the total fault value of the to-be-detected electric energy meter according to the characteristic features after feature enhancement processing, the preset balance parameter, and the total number of fault categories, can be executed in the order of first executing S401 and then executing S402, or in the order of first executing S402 and then executing S401, and the embodiment does not limit this.

[0090] S403 takes the ratio between the initial fault value and the total fault value as the initial fault probability of the to-be-detected electric energy meter in the fault category.

[0091] In some embodiments, the ratio between the initial fault value and the total fault value is taken as the initial fault probability of the to-be-detected electric energy meter in the fault category by the following formula (1-2).

[0092] (1-2)

[0093] wherein L represents the number of layers of the output layer in the fault detection model (usually the last layer), and K represents the total number of fault categories. represents the characteristic feature in the kth dimension, represents the probability that the input characteristic feature belongs to the kth fault category. T is a preset balance parameter. When T > 1, the initial fault probability distribution is more gentle, and the probability difference of each fault category is reduced. When T < 1, the initial fault probability distribution is more sharp, and the probability difference of each fault category is increased. Considering the problem that the to-be-detected electric energy meter in the embodiment has differences, T is about 0.8.

[0094] In the embodiment, based on the output layer in the fault detection model, the initial fault value of the to-be-detected electric energy meter in the fault category is determined according to the characteristic features after the feature enhancement processing and the preset balance parameter; and the fault total value of the to-be-detected electric energy meter is determined according to the characteristic features after the feature enhancement processing, the preset balance parameter and the total number of fault categories; and the ratio between the initial fault value and the fault total value is taken as the initial fault probability of the to-be-detected electric energy meter in the fault category. The embodiment can more accurately determine the initial fault probability of the to-be-detected electric energy meter in different fault categories through the fault detection model.

[0095] Figure 5A FIG. 1 is a flowchart of a process for determining the fault diagnosis result in one embodiment, which specifically includes the following steps:

[0096] S501 For each fault category, the category weight of the fault category is determined according to the confidence of the fault category to the to-be-detected electric energy meter.

[0097] In some embodiments, for each fault category, the category weight of the fault category is determined according to the confidence of the fault category to the to-be-detected electric energy meter through the following formulas (1-3).

[0098] (1-3)

[0099] wherein, represents the confidence of the cth fault category to the to-be-detected electric energy meter, which can usually take the maximum value of the probability of outputting the initial fault probability in the fault category, and C is the number of fault categories.

[0100] S502 The initial fault probability in each fault category, the historical probability distribution and the category weight of the corresponding fault category are weighted to obtain the fault diagnosis result of the to-be-detected electric energy meter.

[0101] In some embodiments, the initial fault probability in each fault category, the historical probability distribution and the category weight of the corresponding fault category are weighted to obtain the fault diagnosis result of the to-be-detected electric energy meter through the following formula (1-4).

[0102] ​(1-4)

[0103] wherein, represents the fault diagnosis result, represents the class weight of the cth fault category (the first fault category is replaced by the historical probability distribution corresponding to the faulty component and / or the preset probability distribution set by the expert, the class weight is allocated by the fault category corresponding classifier performance, and satisfies , C is the total number of fault categories), represents the initial fault probability output under the cth fault category. It should be noted that the first fault category can be determined by the historical probability distribution corresponding to the faulty component, or by the preset probability distribution set by the expert for the faulty component, or by the historical probability distribution and the preset probability distribution together, and the specific implementation mode is not limited in this embodiment.

[0104] It should be noted that the fault diagnosis result in this embodiment represents the probability distribution of the faulty component under different fault categories, so in the iteration process of determining the fault diagnosis result of the current electric energy meter, the fault diagnosis result determined in the last iteration process can be used as a new historical probability distribution, and the historical probability distribution can be used to assist in determining the first fault category in the current iteration process.

[0105] For example, as shown in the fault diagnosis result diagram, Figure 5B In this embodiment, 100 faulty electric energy meters are selected as test samples, including 9 component faults, and 9 component faulty electric energy meters such as liquid crystal screen, transformer, clock battery, capacitor-power supply, thermistor, voltage stabilizing chip-metering, power supply chip, optocoupler-metering, and metering chip are identified. The number of correct identifications is 83, and the identification rate is 83%. Among them, the traditional appearance analysis can only identify the liquid crystal screen and the clock battery, and the present application also has a high identification rate for such components, and can effectively identify other faulty components.

[0106] In this embodiment, for each fault category, the class weight of the fault category is determined according to the confidence of the fault category to the electric energy meter to be detected. The initial fault probability under each fault category and the class weight of the corresponding fault category are weighted to obtain the fault diagnosis result of the electric energy meter to be detected, and the fault diagnosis result can be more accurately determined.

[0107] It should be understood that although the steps in the flowcharts involved in the embodiments described above are shown in sequence according to the arrows, the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, the execution of the steps is not strictly limited in sequence, and the steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the embodiments described above can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of the steps or stages is not necessarily sequential, but can be alternately or alternately executed with at least part of other steps or steps or stages in other steps.

[0108] Based on the same inventive concept, the embodiments of the present application also provide an electric energy meter fault diagnosis device for implementing the above-mentioned electric energy meter fault diagnosis method based on multi-dimensional characteristics. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more electric energy meter fault diagnosis device embodiments provided below can refer to the limitations of the electric energy meter fault diagnosis method described above, which will not be repeated here.

[0109] In an exemplary embodiment, as shown in Figure 6 An electric energy meter fault diagnosis device is provided, comprising: an acquisition module 10, an enhancement module 11, an identification module 12 and a diagnosis module 13, wherein:

[0110] The acquisition module 10 is configured to acquire electric energy meter characteristics of the electric energy meter to be detected in at least one dimension; the electric energy meter characteristics include impedance characteristics, electric quantity characteristics, shape characteristics, general performance characteristics and running state characteristics;

[0111] The enhancement module 11 is configured to perform feature extraction on the electric energy meter characteristics in each dimension respectively, to obtain characteristic features of the electric energy meter to be detected in the corresponding dimension;

[0112] The identification module 12 is configured to perform fault detection on the electric energy meter to be detected according to the characteristic features of the electric energy meter to be detected in at least one dimension through a pre-trained fault detection model, to obtain initial fault probabilities of the electric energy meter to be detected in at least one fault category; the fault detection model is trained based on a multi-layer perception neural network;

[0113] The diagnosis module 13 is configured to determine a fault diagnosis result of the electric energy meter to be detected according to the initial fault probabilities in at least one fault category, a historical probability distribution and a category weight of the corresponding fault category.

[0114] In some embodiments, the recognition module 12 is further configured to perform feature enhancement processing on each characteristic feature according to the characteristic features of the to-be-detected electric energy meter in at least one dimension, to obtain a corresponding characteristic feature after feature enhancement processing; and for each fault category, determine an initial fault probability of the to-be-detected electric energy meter in the fault category according to each characteristic feature after feature enhancement processing, the preset balance parameter, and the total number of fault categories.

[0115] In some embodiments, the recognition module 12 is further configured to determine average features of the to-be-detected electric energy meter according to the characteristic features of the to-be-detected electric energy meter in at least one dimension; and for each characteristic feature in each dimension, perform feature enhancement processing on the characteristic feature according to the characteristic feature and the corresponding average feature, to obtain a corresponding characteristic feature after feature enhancement processing.

[0116] In some embodiments, the recognition module 12 is further configured to determine, based on an output layer in the fault detection model, an initial fault value of the to-be-detected electric energy meter in the fault category according to each characteristic feature after feature enhancement processing and the preset balance parameter; determine a total fault value of the to-be-detected electric energy meter according to each characteristic feature after feature enhancement processing, the preset balance parameter, and the total number of fault categories; and take a ratio between the initial fault value and the total fault value as the initial fault probability of the to-be-detected electric energy meter in the fault category.

[0117] In some embodiments, the diagnosis module 13 is further configured to, for each fault category, determine a category weight of the fault category according to the confidence of the fault category to the to-be-detected electric energy meter; and perform weighted processing on the initial fault probability in each fault category, the historical probability distribution, and the category weight of the corresponding fault category, to obtain a fault diagnosis result of the to-be-detected electric energy meter.

[0118] In some embodiments, the electric energy meter fault diagnosis apparatus further comprises a visualization module configured to display, in a visualization interface, an electric energy meter appearance of the to-be-detected electric energy meter, conventional performance detection data, a running state word, an impedance Bode diagram, an electric quantity grayscale diagram, and a fault diagnosis result; wherein the impedance Bode diagram is converted based on impedance characteristics of the to-be-detected electric energy meter; the electric quantity grayscale diagram is converted based on electric quantity characteristics of the to-be-detected electric energy meter; the electric energy meter appearance is converted based on appearance characteristics of the to-be-detected electric energy meter; the conventional performance detection data is converted based on conventional performance characteristics of the to-be-detected electric energy meter; and the running state word is converted based on running state characteristics of the to-be-detected electric energy meter.

[0119] Each module in the above electric energy meter fault diagnosis apparatus can be realized by software, hardware, or a combination thereof, in whole or in part. Each module can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in the computer device in software form, so as to be called and executed by a processor to perform operations corresponding to each module.

[0120] In an example embodiment, a computer device, which can be a server, is provided, and an internal structure diagram of the computer device can be as shown in Figure 7 The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store data. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through a network connection. The computer program is executed by the processor to implement a power meter fault diagnosis method.

[0121] Those skilled in the art can understand that Figure 7 The structure shown in the above embodiment is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the diagram, or combine certain components, or have a different arrangement of components.

[0122] In an example embodiment, a computer device is also provided, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0123] In an example embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the steps in the above method embodiments.

[0124] In an example embodiment, a computer program product is provided, which includes a computer program. The computer program is executed by a processor to implement the steps in the above method embodiments.

[0125] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use, and processing of related data need to comply with relevant regulations.

[0126] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0127] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0128] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A power meter fault diagnosis method based on multi-dimensional characteristics, characterized in that, The method comprises: obtaining an electric energy meter characteristic of a to-be-detected electric energy meter in at least one dimension; the electric energy meter characteristic comprises an impedance characteristic, an electric quantity characteristic, an appearance characteristic, a general performance characteristic, and an operating state characteristic; performing feature extraction on the electric energy meter characteristic in each dimension respectively to obtain a characteristic feature of the to-be-detected electric energy meter in the corresponding dimension; performing feature enhancement processing on each characteristic feature according to the characteristic features of the to-be-detected electric energy meter in at least one dimension to obtain a corresponding characteristic feature after feature enhancement processing; for each fault category, determining an initial fault value of the to-be-detected electric energy meter in the fault category based on an output layer in a fault detection model, each characteristic feature after feature enhancement processing, and a preset balance parameter; the fault detection model is obtained based on a multilayer perceptron neural network; and determining a total fault value of the to-be-detected electric energy meter according to each characteristic feature after feature enhancement processing, a preset balance parameter, and a total number of fault categories; taking a ratio between the initial fault value and the total fault value as an initial fault probability of the to-be-detected electric energy meter in the fault category; determining a fault diagnosis result of the to-be-detected electric energy meter according to the initial fault probability in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category; wherein the initial fault probability is determined based on the following formula: wherein L represents the number of layers of the output layer in the fault detection model, K represents the total number of fault categories, represents a characteristic feature in the kth dimension, represents a characteristic feature in the k'th dimension, represents the probability that the input characteristic feature belongs to the kth fault category, and T is a preset balance parameter.

2. The method of claim 1, wherein, the feature enhancement processing on each characteristic feature according to the characteristic features of the to-be-detected electric energy meter in at least one dimension to obtain a corresponding characteristic feature after feature enhancement processing comprises: determining an average feature of the to-be-detected electric energy meter according to the characteristic features of the to-be-detected electric energy meter in at least one dimension; for each characteristic feature in each dimension, performing feature enhancement processing on the characteristic feature according to the characteristic feature and a corresponding average feature to obtain a corresponding characteristic feature after feature enhancement processing.

3. The method of claim 1, wherein, the determination of the fault diagnosis result of the to-be-detected electric energy meter according to the initial fault probability in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category comprises: for each fault category, determining a category weight of the fault category according to a confidence degree of the fault category on the to-be-detected electric energy meter; performing weighted processing on the initial fault probability in each fault category, a historical probability distribution, and a category weight of the corresponding fault category to obtain the fault diagnosis result of the to-be-detected electric energy meter.

4. The method according to any one of claims 1 to 3, characterized in that, The method further comprises: displaying an electric energy meter appearance, general performance detection data, an operating state word, an impedance Bode diagram, an electric quantity grayscale diagram, and a fault diagnosis result of the to-be-detected electric energy meter in a visual interface; wherein the impedance Bode diagram is converted based on the impedance characteristic of the to-be-detected electric energy meter; the electric quantity grayscale diagram is converted based on the electric quantity characteristic of the to-be-detected electric energy meter; the electric energy meter appearance is converted based on the appearance characteristic of the to-be-detected electric energy meter; the general performance detection data is converted based on the general performance characteristic of the to-be-detected electric energy meter; and the operating state word is converted based on the operating state characteristic of the to-be-detected electric energy meter.

5. A power meter fault diagnosis device based on multi-dimensional characteristics, characterized in that, The device comprises: An acquisition module is configured to acquire an electric energy meter characteristic of a to-be-detected electric energy meter in at least one dimension, wherein the electric energy meter characteristic includes an impedance characteristic, an electric quantity characteristic, an external shape characteristic, a general performance characteristic, and an operating state characteristic. An enhancement module is configured to perform feature extraction on the electric energy meter characteristic in each dimension respectively, to obtain a characteristic feature of the to-be-detected electric energy meter in the corresponding dimension. An identification module is configured to perform feature enhancement processing on each characteristic feature according to the characteristic feature of the to-be-detected electric energy meter in at least one dimension, to obtain a characteristic feature after feature enhancement processing; for each fault category, based on an output layer in a fault detection model, determine an initial fault value of the to-be-detected electric energy meter in the fault category according to each characteristic feature after feature enhancement processing and a preset balance parameter; the fault detection model is obtained based on a multilayer perception neural network; and determine a total fault value of the to-be-detected electric energy meter according to each characteristic feature after feature enhancement processing, a preset balance parameter, and a total number of fault categories; take a ratio between the initial fault value and the total fault value as an initial fault probability of the to-be-detected electric energy meter in the fault category; wherein the initial fault probability is determined based on the following formula: wherein L represents the number of layers of the output layer in the fault detection model, K represents the total number of fault categories, represents a characteristic feature in the kth dimension, represents a characteristic feature in the k'th dimension, represents the probability that the input characteristic feature belongs to the kth fault category, and T is a preset balance parameter; A diagnosis module is configured to determine a fault diagnosis result of the to-be-detected electric energy meter according to the initial fault probability in at least one fault category, a historical probability distribution, and a category weight of the corresponding fault category. 6.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-5 when the computer program is executed by the processor. The processor executes the computer program to realize the steps of the method in any one of claims 1 to 4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method in any one of claims 1 to 4.

8. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method in any one of claims 1 to 4. The computer program is executed by the processor to realize the steps of the method in any one of claims 1 to 4.

Images