Clustering analysis method and system for meter burning fault of disassembled electric energy meter
By adopting cluster analysis methods in the detection of power meter faults, eliminating wrong values, filling missing data and adjusting weight factors, efficient detection of power meter burning faults is achieved, and the problems of inefficiency and high cost in the existing technology are solved. Ensure the stable operation of the power system.
Patent Information
- Application Number
- CN202410991873.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-11-15
AI Technical Summary
In the prior art, the detection of electric energy meter burning faults depends on manual inspection and maintenance, which is inefficient and expensive, making it difficult to achieve in large-scale applications.
The clustering analysis method of disassemblying the power meter burning fault is adopted. By connecting the power meter into the test circuit, characteristic data is collected, local outliers are used to eliminate error values, multivariate interpolation model fills the missing data, and normalized weight factors are introduced. And use the mean clustering algorithm to perform fault analysis.
It improves the efficiency and accuracy of power meter fault detection, reduces detection deviation, and provides an efficient solution suitable for large-scale applications to ensure the stable operation of the power system.
Smart Images

Figure CN118965039A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of electric power systems, and in particular to a cluster analysis method and system for a dismantled electric energy meter burnout fault. Background Art
[0002] As the core equipment for electricity trading and settlement, the energy meter is installed after the user's incoming line switch and before the power-consuming equipment. Its failure may lead to multiple levels of adverse effects. Once a failure occurs, the most direct consequence is power outage for the user, which may affect the user's daily electricity experience at the least, or may cause significant economic and property losses to the user at the worst. Especially in the case of modern society's high dependence on electricity, the stable operation of the energy meter is directly related to the normal production and living order of residents, enterprises and public institutions. Among the many energy meter failures, the burnout failure is a particularly common one, and its root causes involve multiple aspects such as untightened terminal blocks, poor contact, and excessive load. This failure may not only lead to a decline in the performance of the energy meter, but may also have a serious chain effect on the normal operation of the entire power system. Therefore, it is crucial to detect and quickly resolve the energy meter burnout failure in advance to ensure the stable operation of the power system.
[0003] In current technology applications, the solution to electric energy meter burnout failures usually relies on manual detection and repair. However, this traditional method has obvious drawbacks, including low efficiency, high cost, and difficulty in large-scale application.
[0004] In recent years, with the rapid development of cluster analysis technology, it has made significant progress in the field of fault diagnosis. Cluster analysis technology stands out for its efficient processing ability of large-scale data. By automatically grouping similar data points, it helps to deeply understand the mechanism of electric energy meter burnout failure. Summary of the invention
[0005] The technical problem to be solved by the present invention is as follows: In view of the above-mentioned problems in the prior art, a method and system for clustering analysis of burnt meter failures of dismantled electric energy meters are provided, which provides a new and efficient solution for determining electric energy meters with similar failure modes, and provides a reliable basis for subsequent in-depth analysis and improvement.
[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:
[0007] A cluster analysis method for a dismantled electric energy meter burnout fault comprises the following steps:
[0008] Connect each electric energy meter to the test circuit for experiment and collect corresponding characteristic data to obtain the original characteristic matrix;
[0009] Use the local outlier factor to eliminate the erroneous values of each type of feature in the original feature matrix to obtain the first feature matrix;
[0010] Use the multivariate interpolation model to interpolate the missing data of each type of feature in the first feature matrix in turn to obtain the second feature matrix;
[0011] A normalized weight factor is introduced into the second characteristic matrix to obtain a third characteristic matrix;
[0012] The mean value clustering algorithm is used to perform cluster analysis on the data in the third characteristic matrix, and the electric energy meter with the burnout fault is determined according to the cluster analysis result.
[0013] Furthermore, the test circuit includes a switch before the meter, an emergency stop switch, an incoming line switch, a switch after the meter, and an adjustable load. When each electric energy meter is connected to the test circuit for an experiment, the following steps are performed:
[0014] Connect the live wire incoming switch and the neutral wire incoming switch of the energy meter to the external power supply through the switch in front of the meter, the emergency stop switch and the incoming switch in sequence, and connect the live wire outgoing switch and the neutral wire outgoing switch of the energy meter to the adjustable load through the switch behind the meter;
[0015] Change the wire material of the test circuit and adjust the tightness of the terminal blocks to collect characteristic information of the electric energy meter at different time nodes.
[0016] Furthermore, the original feature matrix expression is as follows:
[0017]
[0018] Among them, X1 is the electrical parameter characteristics of the electric energy meter, including current and voltage; X2 is the environmental factor characteristics of the electric energy meter, including temperature and humidity; X3 is the mechanical state characteristics of the electric energy meter, including terminal deformation, screw tightening degree, and number of wire strands, which are represented by label codes.
[0019] Furthermore, the local outlier factor is used to eliminate the erroneous values of each type of feature in the original feature matrix, including:
[0020] Calculate feature X n The local outlier factor of the feature of each point in . If the local outlier factor is greater than the preset threshold, the corresponding point is marked as a null value.
[0021] Furthermore, we calculate the feature X n When the local outlier factor of the feature of each point in is expressed as follows:
[0022]
[0023] Among them, x n (i) is feature Xn The feature of the i-th point, N k (x n (i)) is feature X n The kth distance neighborhood of the feature of the i-th point, lrd k (x n (i)) represents feature X n The local k-th density of the feature at the i-th point, o is the domain point.
[0024] Furthermore, when the missing data of each type of feature in the first feature matrix is interpolated in sequence using the multivariate interpolation model, it includes:
[0025] Establish a multivariate interpolation model corresponding to each type of feature, the expression is as follows:
[0026]
[0027] in, In the tth iteration, feature X j The current iteration requires the prediction value of the interpolated feature, β0 represents the intercept term of the model, β k represents the model coefficient of the kth feature, Represents other features X k For features that need to be interpolated The error term ∈ represents the error between the model prediction and the actual value;
[0028] The predicted value of the missing value position of the corresponding feature is calculated according to the multivariate interpolation model, the predicted value is filled into the corresponding missing value position, and then the predicted value of the next missing value position of the corresponding feature is calculated according to the multivariate interpolation model until all missing values of the corresponding feature are processed.
[0029] Furthermore, the first characteristic matrix includes the mechanical state characteristics of the electric energy meter, and the multivariate interpolation model expression corresponding to the mechanical state characteristics is as follows:
[0030]
[0031] in, Indicates that in the tth iteration, the feature The probability of taking the value of 1, β0 represents the intercept term of the model, β k represents the model coefficient of the kth feature, Represents other features X k For features that need to be interpolated degree of impact.
[0032] Furthermore, the second characteristic matrix includes the electrical parameter characteristics, environmental factor characteristics and mechanical state characteristics of the electric energy meter. When a normalized weight factor is introduced into the second characteristic matrix, it includes:
[0033] The dynamic weight factors corresponding to the electrical parameter characteristics and the environmental factor characteristics are calculated respectively, and then the dynamic weight factors corresponding to the electrical parameter characteristics and the dynamic weight factors corresponding to the environmental factor characteristics are normalized.
[0034] Furthermore, the dynamic weight factor expressions corresponding to the electrical parameter characteristics and environmental factor characteristics are as follows:
[0035]
[0036] Among them, α is the dynamic weight factor of the electrical parameter feature, β is the dynamic weight factor of the environmental factor feature, x1(i) is the feature of the i-th point of the electrical parameter feature, and x2(i) is the feature of the i-th point of the environmental factor feature.
[0037] The present invention also proposes a cluster analysis system for disassembled electric energy meter burnout faults, comprising a microprocessor and a computer-readable storage medium connected to each other, wherein the microprocessor is programmed or configured to execute any one of the disassembled electric energy meter burnout fault cluster analysis methods.
[0038] Compared with the prior art, the advantages of the present invention are:
[0039] The present invention uses a multivariate interpolation method based on chain equations to interpolate the missing data in each feature in turn, and gradually optimizes the missing values in the characteristic data of the electric energy meter through multiple iterations, thereby reducing deviations and improving data utilization and clustering accuracy, making the time series characteristics of the electric energy meter more reliable, being able to handle the missing of multiple characteristic variables, effectively completing the missing data of the electric energy meter, and maintaining the internal structure and mutual relationships of the data.
[0040] The present invention takes into account that the importance of various features in different working stages of the electric energy meter may change, introduces a normalized weight factor, completes the dynamic adjustment of the importance of the characteristic data of the electric energy meter, and adaptively adjusts the weight factor to adapt to the working characteristics of the electric energy meter in different stages, which can more accurately reflect the importance of the features under different conditions, and thus is more adaptable. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention.
[0042] Figure 2 The figure is a wiring diagram of an electric energy meter of a test circuit in an embodiment of the present invention.
[0043] Figure 3 This is a clustering result diagram of an embodiment of the present invention. DETAILED DESCRIPTION
[0044] The present invention is further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.
[0045] Embodiment 1
[0046] This embodiment proposes a clustering analysis method for the burned-out meter fault of a dismantled electric energy meter, introduces a local outlier factor (LOF) to eliminate erroneous values in the original data; uses a multivariate interpolation method based on chain equations to interpolate the missing data in each feature in turn, and gradually optimizes the missing values in the characteristic data of the electric energy meter through multiple iterations, so that the time series characteristics of the electric energy meter are more reliable; considering that the importance of various features may change in different working stages of the electric energy meter, a normalized weight factor is introduced to achieve dynamic adjustment of the importance of the characteristic data of the electric energy meter; finally, a mean clustering algorithm is used to complete the clustering analysis of the burned-out meter fault of the dismantled electric energy meter.
[0047] like Figure 1 As shown, the method of this embodiment includes the following steps:
[0048] S101) connecting each electric energy meter to a test circuit for experiment and collecting corresponding characteristic data to obtain an original characteristic matrix;
[0049] S102) using a local outlier factor to eliminate erroneous values of each type of feature in the original feature matrix. In this embodiment, the feature matrix after eliminating the erroneous values is called a first feature matrix for distinction;
[0050] S103) using a multivariate interpolation model to interpolate the missing data of each type of feature in the first feature matrix in turn, and gradually optimize the missing values in the feature data of the electric energy meter through multiple iterations. In this embodiment, the feature matrix after interpolation is referred to as a second feature matrix for distinction;
[0051] S104) introducing a normalized weight factor into the second characteristic matrix to complete the dynamic adjustment of the importance of the characteristic data of the electric energy meter. In this embodiment, the characteristic matrix after the normalized weight factor is introduced is called the third characteristic matrix for distinction;
[0052] S105) Using a mean value clustering algorithm to perform cluster analysis on the data in the third characteristic matrix, and determining the electric energy meter with a burnout fault according to the cluster analysis result.
[0053] Each step is described below.
[0054] In step S101 of this embodiment, the test circuit is as follows: Figure 2 As shown in the figure, the existing internal wiring of the meter box is removed and the existing wires are cut, including the switch before the meter, the emergency stop switch, the incoming line switch, the switch after the meter and the adjustable load. Figure 2 In the diagram, ports 1 / 2 / 3 / 4 represent the live incoming switch, live outgoing switch, neutral incoming switch, and neutral outgoing switch of the single-phase energy meter. When each energy meter is connected to the test circuit for experiment, it includes:
[0055] Connect the live wire incoming switch and the neutral wire incoming switch of the energy meter to the external power supply through the switch in front of the meter, the emergency stop switch and the incoming switch in sequence, and connect the live wire outgoing switch and the neutral wire outgoing switch of the energy meter to the adjustable load through the switch behind the meter;
[0056] Change the wire material of the test circuit and adjust the tightness of the terminal blocks to collect characteristic information of the electric energy meter at different time nodes.
[0057] In this example, four types of aluminum wire were selected: 4mm 2 Single strand, 6mm 2 Single strand, 10mm 2 Multi-strand, 16mm 2 Multi-strand; 5 types of copper wire Wire material: 4mm 2 Single strand, 4mm 2 Multi-strand, 6mm 2 Single strand, 6mm 2 Multi-strand, 10mm 2 Multiple strands, copper-aluminum transition joints are also prepared. Use different wire materials and wiring methods, and adjust the tightening degree of the screws to connect the energy meter to the Figure 2 The test circuit is used to collect the characteristic information of electrical parameters such as current and voltage at different time nodes of the electric energy meter, and at the same time, observe the deformation of the terminal block. "Loosening" is defined as the screw just touching the wiring, "tightening" is defined as the screw just touching the wiring and then tightening it half a turn to one turn, and "tightening" is defined as the state between the two states. The electric meter feature data set finally obtained after the experiment contains seven input features: voltage, temperature, terminal deformation, degree of tightening of screws, normal working time, wire condition, and whether copper-aluminum transition joints are used. The original feature matrix expression obtained is as follows:
[0058]
[0059] Among them, X1 is the electrical parameter characteristics of the electric energy meter, including current and voltage; X2 is the environmental factor characteristics of the electric energy meter, including temperature and humidity; X3 is the mechanical state characteristics of the electric energy meter, including terminal deformation, screw tightening degree, and number of wire strands, which are represented by label codes.
[0060] In order to reduce the negative impact of erroneous data in the feature matrix X on the anomaly detection effect, this embodiment uses a local outlier factor (LOF) to eliminate erroneous values in the feature X, n∈1,…,N in step S102. When the local outlier factor is used to eliminate the erroneous values of each type of feature in the original feature matrix, it includes:
[0061] Calculate feature X n The local outlier factor of the feature of each point in is expressed as follows:
[0062]
[0063] Among them, x n (i) is feature X n The feature of the i-th point, N k( x n (i) ) For feature X n The kth distance neighborhood of the feature of the i-th point, lrd k (x n (i)) represents feature X n The local k-th density of the feature at the i-th point, o is the domain point;
[0064] If the local outlier factor is greater than the preset threshold, the corresponding point is an outlier. After the local outlier processing, the original feature data of the electric energy meter is converted into feature X n The outliers in are marked with the null value NaN.
[0065] Missing data is a common problem in cluster analysis research. If not handled properly, it may bias the clustering results. Multivariate interpolation is considered to be a reliable missing value estimation method. It takes into account the uncertainty of the interpolated values, solves the problem of underestimated standard errors, and obtains missing value estimates with smaller deviations. In the missing data filling step of the electric energy meter, these issues should be considered: 1) Consider the process that causes missing values; 2) Preserve the relationship between features; 3) Consider the uncertainty of these relationships.
[0066] Step S103 of this embodiment is based on multivariate interpolation of chain equations. By specifying a model for each type of incomplete feature, the missing values in the original data are gradually optimized through multiple iterations. This method has significant advantages in processing missing feature data of electric energy meters. First, it can process missing data of multiple feature variables, ensuring that the interpolated data maintains the original internal structure and mutual relationship, thereby improving the accuracy of interpolation. Secondly, chain equation multivariate interpolation allows the use of multiple models, can effectively process numerical and state data, reduce deviations through multiple iterations, and improve data utilization and clustering accuracy. In addition, the randomness introduced in the interpolation process can reflect the uncertainty of the data, helping to more comprehensively understand the range of variation and potential risks of electric energy meter data.
[0067] Therefore, when the missing data of each type of feature in the first feature matrix are interpolated in turn using the multivariate interpolation model, it includes:
[0068] 1) Establish a multivariate interpolation model corresponding to each type of feature. Assume that we are in the tth round of iteration and the data after the tth round of interpolation is X (t) , then the multivariate imputation model is defined as follows:
[0069]
[0070] in, In the tth iteration, feature X j The current iteration requires the prediction value of the interpolated feature, β0 represents the intercept term of the model, β k represents the model coefficient of the kth feature, Represents other features X k For features that need to be interpolated The error term ∈ represents the error between the model prediction and the actual value;
[0071] For the mechanical state characteristics of the electric energy meter, the corresponding multivariate interpolation model expression is as follows:
[0072]
[0073] in, Indicates that in the tth iteration, the feature The probability of taking the value of 1, β0 represents the intercept term of the model, β k represents the model coefficient of the kth feature, Represents other features X k For features that need to be interpolated the extent of the impact;
[0074] 2) Calculate the predicted value of the missing value position of the corresponding feature according to the multivariate interpolation model, and fill the predicted value into the corresponding missing value position;
[0075] 3) Repeat the above steps to calculate the predicted value of the next missing value position of the corresponding feature according to the multivariate interpolation model, so as to interpolate the missing data of each feature in sequence until all missing values of the corresponding feature are processed, and output the final interpolated feature matrix X.
[0076] In step S104 of this embodiment, the importance of various features may change in different working stages, and the weight factor is adaptively adjusted according to the changes in the working state and environmental conditions of the electric energy meter.
[0077] When the load level of the energy meter changes greatly, the electrical parameter characteristics may be more critical, so their weight is increased at this stage; when the environmental conditions change and the temperature or humidity fluctuates greatly, the weight of the environmental factors may need to be adjusted accordingly. Therefore, when the normalized weight factors are introduced in the second feature matrix, they include:
[0078] The dynamic weight factors corresponding to the electrical parameter characteristics and environmental factor characteristics are calculated respectively. The expressions of the dynamic weight factors corresponding to the electrical parameter characteristics and environmental factor characteristics are as follows:
[0079]
[0080] Among them, α is the dynamic weight factor of the electrical parameter feature, β is the dynamic weight factor of the environmental factor feature, x1(i) is the feature of the i-th point of the electrical parameter feature, and x2(i) is the feature of the i-th point of the environmental factor feature;
[0081] Next, in order to ensure the effectiveness of the dynamic weights, so that the system can adjust the relative importance of the features according to the state of the electric energy meter without being affected by the feature weight scale, the dynamic weight factors need to be normalized. The formulas for normalizing the dynamic weight factors corresponding to the electrical parameter features and the dynamic weight factors corresponding to the environmental factor features are as follows:
[0082]
[0083] The characteristic matrix of the electric energy meter that finally introduces the normalized weight factor, that is, the third characteristic matrix, is:
[0084]
[0085] Among them, X1 is the electrical parameter characteristic after the erroneous values are eliminated by local outlier factors and the missing data are interpolated by the multivariate interpolation model; X2 is the environmental factor characteristic after the erroneous values are eliminated by local outlier factors and the missing data are interpolated by the multivariate interpolation model; X3 is the mechanical state characteristic after the erroneous values are eliminated by local outlier factors and the missing data are interpolated by the multivariate interpolation model.
[0086] In step S105 of this embodiment, when the mean clustering algorithm is used to perform cluster analysis on the data in the third characteristic matrix, the third characteristic matrix is input into the k-Means mean clustering algorithm, and k=2 is set to divide the data into two categories, one is normal operation and the other is a burnout fault, and the clustering analysis of the electric energy meter burnout fault is completed. Among them, the Euclidean distance calculation formula of the k-Means mean clustering algorithm is:
[0087]
[0088] In the above formula, x(i) is the feature data, there are N of them, k j (j=1 or j=2), representing two cluster centers. Figure 3 The clustering results are shown in Figure 1. The red dots represent the cluster centers of each category. Using external labels to verify, we can see that the yellow dots represent burnout and the purple dots represent non-burnout. At the same time, by checking the original features of the two categories of data, we found that when other conditions are the same, the smaller the wire diameter, the greater the resistance, and the greater the probability of burnout; considering the corrosion resistance of the wire, the multi-strand wire has better conductivity after many years, so the multi-strand wire under the same conditions has a lower probability of burnout than the single-strand wire.
[0089] Embodiment 2
[0090] This embodiment proposes a cluster analysis system for disassembled electric energy meter burnout faults, including a microprocessor and a computer-readable storage medium connected to each other, wherein the microprocessor is programmed or configured to execute the cluster analysis method for disassembled electric energy meter burnout faults described in the first embodiment.
[0091] This embodiment further proposes a computer-readable storage medium, in which a computer program is stored. The computer program is used to be programmed or configured by a microprocessor to execute the clustering analysis method for the burnt meter failure of the dismantled electric energy meter described in the first embodiment.
[0092] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0093] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.
Claims
1. A cluster analysis method for the failure of a dismantled electric energy meter to burn out, characterized in that: The following steps are involved: Connect each electric energy meter to the test circuit for experiment and collect corresponding characteristic data to obtain the original characteristic matrix; Use the local outlier factor to eliminate the erroneous values of each type of feature in the original feature matrix to obtain the first feature matrix; Use the multivariate interpolation model to interpolate the missing data of each type of feature in the first feature matrix in turn to obtain the second feature matrix; A normalized weight factor is introduced into the second characteristic matrix to obtain a third characteristic matrix; The mean value clustering algorithm is used to perform cluster analysis on the data in the third characteristic matrix, and the electric energy meter with the burnout fault is determined according to the cluster analysis result.
2. The cluster analysis method for the burnt meter failure of the dismantled electric energy meter according to claim 1 is characterized in that: The test circuit includes a switch before the meter, an emergency stop switch, an incoming line switch, a switch after the meter, and an adjustable load. When each electric energy meter is connected to the test circuit for an experiment, the following steps are performed: Connect the live wire incoming switch and the neutral wire incoming switch of the energy meter to the external power supply through the switch in front of the meter, the emergency stop switch and the incoming switch in sequence, and connect the live wire outgoing switch and the neutral wire outgoing switch of the energy meter to the adjustable load through the switch behind the meter; Change the wire material of the test circuit and adjust the tightness of the terminal blocks to collect characteristic information of the electric energy meter at different time nodes.
3. The cluster analysis method for the burnt meter failure of the dismantled electric energy meter according to claim 1 is characterized in that: The original feature matrix expression is as follows: Among them, X1 is the electrical parameter characteristics of the electric energy meter, including current and voltage; X2 is the environmental factor characteristics of the electric energy meter, including temperature and humidity; X3 is the mechanical state characteristics of the electric energy meter, including terminal deformation, screw tightening degree, and number of wire strands, which are represented by label codes.
4. The cluster analysis method for the burnt meter failure of the dismantled electric energy meter according to claim 1 is characterized in that: When using the local outlier factor to eliminate the erroneous values of each type of feature in the original feature matrix, including: Calculate feature X n The local outlier factor of the feature of each point in . If the local outlier factor is greater than the preset threshold, the corresponding point is marked as a null value.
5. The cluster analysis method for the burnt meter failure of the dismantled electric energy meter according to claim 4 is characterized in that: Calculate feature X n When the local outlier factor of the feature of each point in is expressed as follows: Among them, x n (i) is feature X n The feature of the i-th point, N k (x n (i)) is feature X n The kth distance neighborhood of the feature of the i-th point, lrd k (x n (i)) represents feature X n The local k-th density of the feature at the i-th point, o is the domain point.
6. The cluster analysis method for the burnt meter failure of the dismantled electric energy meter according to claim 1 is characterized in that: When using the multivariate interpolation model to interpolate the missing data of each type of feature in the first feature matrix in turn, it includes: Establish a multivariate interpolation model corresponding to each type of feature, the expression is as follows: in, In the tth iteration, feature X j The current iteration requires the prediction value of the interpolated feature, β0 represents the intercept term of the model, β k represents the model coefficient of the kth feature, Represents other features X k For features that need to be interpolated The error term ∈ represents the error between the model prediction and the actual value; The predicted value of the missing value position of the corresponding feature is calculated according to the multivariate interpolation model, the predicted value is filled into the corresponding missing value position, and then the predicted value of the next missing value position of the corresponding feature is calculated according to the multivariate interpolation model until all missing values of the corresponding feature are processed.
7. The cluster analysis method for the burnt meter failure of the dismantled electric energy meter according to claim 6 is characterized in that: The first characteristic matrix includes the mechanical state characteristics of the electric energy meter. The multivariate interpolation model expression corresponding to the mechanical state characteristics is as follows: in, Indicates that in the tth iteration, the feature The probability of taking the value of 1, β0 represents the intercept term of the model, β k represents the model coefficient of the kth feature, Represents other features X k For features that need to be interpolated degree of impact.
8. The cluster analysis method for the failure of a dismantled electric energy meter to burn out according to claim 1 is characterized in that: The second characteristic matrix includes the electrical parameter characteristics, environmental factor characteristics and mechanical state characteristics of the electric energy meter. When the normalized weight factor is introduced into the second characteristic matrix, it includes: The dynamic weight factors corresponding to the electrical parameter characteristics and the environmental factor characteristics are calculated respectively, and then the dynamic weight factors corresponding to the electrical parameter characteristics and the dynamic weight factors corresponding to the environmental factor characteristics are normalized.
9. The cluster analysis method for the failure of a dismantled electric energy meter to burn out according to claim 8 is characterized in that: The dynamic weight factor expressions corresponding to the electrical parameter characteristics and environmental factor characteristics are as follows: Among them, α is the dynamic weight factor of the electrical parameter feature, β is the dynamic weight factor of the environmental factor feature, x1(i) is the feature of the i-th point of the electrical parameter feature, and x2(i) is the feature of the i-th point of the environmental factor feature.
10. A cluster analysis system for dismantled electric energy meter burnout faults, characterized in that: The method comprises a microprocessor and a computer-readable storage medium connected to each other, wherein the microprocessor is programmed or configured to execute the method for clustering analysis of burnt meter faults of dismantled electric energy meters as claimed in any one of claims 1 to 9.
