Intelligent fault diagnosis method, device, equipment, and storage medium for electricity metering boxes

Abstract

This application provides an intelligent fault diagnosis method, device, equipment, and storage medium for electricity metering boxes, belonging to the field of fault diagnosis technology. The method includes: comparing each feature component in various features with its corresponding preset benchmark features to filter out abnormal feature components, and constructing a suspected fault feature vector based on these abnormal feature components; determining the weight of each suspected fault feature component in the suspected fault feature vector, and determining the fault probability of each suspected fault feature component based on the weight; for suspected fault feature components whose fault probability is less than a corresponding preset probability threshold, extracting the corresponding associated features from the comprehensive feature vector, and determining whether the electricity metering box is faulty based on the associated features. This application can improve the accuracy and reliability of fault diagnosis for electricity metering boxes.

Description

Technical Field

[0001] This application belongs to the field of fault diagnosis technology, and more specifically, it relates to an intelligent fault diagnosis method, device, equipment, and storage medium for electricity metering boxes. Background Technology

[0002] Electricity metering boxes are core devices in smart grids, responsible for electricity metering, data acquisition, and transmission, and are widely used in residential, commercial, and industrial sectors. With the large-scale development of smart grids, the operating environment of electricity metering boxes is becoming increasingly complex. Factors such as lightning strikes, overvoltage, high humidity corrosion, and component aging can easily trigger various faults, leading not only to metering errors and disputes between power suppliers and users, but also potentially causing fault propagation and affecting the safe and stable operation of the power grid. Therefore, accurate and efficient fault diagnosis of electricity metering boxes is crucial.

[0003] In existing technologies, fault diagnosis of electricity metering boxes mostly adopts fixed threshold alarms, single parameter detection, or simple data comparison. Some solutions combine differential current detection and remote data reporting to achieve fault identification. However, the above-mentioned existing technologies have obvious defects. For example, relying on a single parameter or fixed threshold for judgment cannot take into account the correlation of multiple source parameters, and it is easy to misjudge transient interference as a fault or miss progressive faults such as current transformer error drift. This results in a high false alarm and false alarm rate, which cannot meet the high-precision diagnostic requirements of intelligent operation and maintenance of the power grid. Summary of the Invention

[0004] This application provides a method, device, equipment, and storage medium for intelligent fault diagnosis of electricity metering boxes, so as to improve the accuracy and reliability of fault diagnosis of electricity metering boxes.

[0005] According to one aspect of the embodiments of this application, an intelligent fault diagnosis method for an electricity metering box is provided, comprising: The electrical parameters, environmental condition parameters, and box condition parameters of the power metering box are obtained; features are extracted from the electrical parameters, environmental condition parameters, and box condition parameters respectively to obtain multiple types of features, and the multiple types of features are fused to obtain a comprehensive feature vector; Each feature component in each type of feature is compared with its corresponding preset benchmark feature to filter out the feature components that are abnormal, and a suspected fault feature vector is constructed based on the feature components that are abnormal. Determine the weight of each suspected fault feature component in the suspected fault feature vector, and determine the fault probability of each suspected fault feature component based on the weight. The failure probability of each suspected fault feature component is compared with its corresponding preset probability threshold. For suspected fault feature components whose failure probability is less than the corresponding preset probability threshold, the associated features corresponding to the suspected fault feature component are extracted from the comprehensive feature vector. Based on the associated features, it is determined whether the power metering box is faulty.

[0006] According to one aspect of the embodiments of this application, an intelligent fault diagnosis device for an electricity metering box is provided, comprising: The feature extraction module is used to obtain the electrical parameters, environmental status parameters, and box status parameters of the power metering box; it performs feature extraction on the electrical parameters, environmental status parameters, and box status parameters respectively to obtain multiple types of features, and then fuses the multiple types of features to obtain a comprehensive feature vector; The suspected fault determination module is used to compare each feature component in various features with its corresponding preset benchmark features, filter out the feature components that are abnormal, and construct a suspected fault feature vector based on the abnormal feature components. The fault probability determination module is used to determine the weight of each suspected fault feature component in the suspected fault feature vector, and to determine the fault probability of each suspected fault feature component based on the weight. The fault diagnosis module is used to compare the fault probability of each suspected fault feature component with its corresponding preset probability threshold. For suspected fault feature components whose fault probability is less than the corresponding preset probability threshold, the module extracts the associated features corresponding to the suspected fault feature component from the comprehensive feature vector and determines whether the power metering box is faulty based on the associated features.

[0007] According to one aspect of the embodiments of this application, an electronic device is provided, the electronic device including a processor and a memory, the memory storing a computer program, the computer program being loaded and executed by the processor to implement the above-described intelligent fault diagnosis method for electricity metering boxes.

[0008] According to one aspect of the embodiments of this application, the computer program product includes a computer program stored in a computer-readable storage medium. A processor of an electronic device reads the computer program from the computer-readable storage medium and executes the computer program, causing the electronic device to perform the aforementioned intelligent fault diagnosis method for electricity metering boxes.

[0009] The technical solutions provided in this application embodiment may have the following beneficial effects: This application embodiment simultaneously acquires the electrical parameters, environmental status parameters, and box status parameters of the power metering box. Features are extracted from each parameter and then fused to form a comprehensive feature vector, achieving comprehensive integration of multi-source information. This avoids the limitations of single-parameter diagnosis and improves the comprehensiveness of diagnosis from the source. This application embodiment compares various features with preset benchmarks to screen for anomalies and constructs a suspected fault feature vector. Weights are assigned based on feature correlation and anomaly severity, and the fault probability is calculated to achieve quantitative fault judgment, which is more accurate than existing qualitative judgment methods. For low-probability suspected faults, related features are extracted from the comprehensive feature vector for secondary verification, effectively eliminating false positives and further improving diagnostic accuracy, reducing ineffective maintenance due to false positives and fault expansion caused by missed diagnoses. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A flowchart illustrating the intelligent fault diagnosis method for electricity metering boxes provided in this application embodiment; Figure 2 This is a structural block diagram of the intelligent fault diagnosis device for electricity metering boxes provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0013] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0014] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0015] Figure 1 This is a flowchart illustrating the intelligent fault diagnosis method for electricity metering boxes provided in this application embodiment. The method is executed by an electronic device and may include: S101: Obtain the electrical parameters, environmental status parameters, and box status parameters of the power metering box; extract features from the electrical parameters, environmental status parameters, and box status parameters respectively to obtain multiple types of features, and fuse the multiple types of features to obtain a comprehensive feature vector.

[0016] In this embodiment, the electricity metering box is a box used to measure electricity consumption and install electricity metering-related equipment, such as a box used for centralized electricity metering in residential areas or industrial parks. Electrical parameters are parameters reflecting the electrical operating status of the electricity metering box, such as voltage and current. Environmental state parameters are parameters reflecting the environmental conditions of the electricity metering box, such as internal temperature and humidity. Box state parameters are parameters reflecting the structural operating status of the electricity metering box itself, such as the open / closed state of the box door. Feature extraction is the process of processing various parameters to extract key information that can be used for subsequent analysis. Multiple features are the various key information obtained after feature extraction. Fusion is the process of integrating multiple features into a unified data carrier. The comprehensive feature vector is the unified data form containing multiple feature information obtained after fusion.

[0017] Considering that power metering box malfunctions are affected by multiple factors such as electrical operation, environmental conditions, and the condition of the box itself, a single parameter cannot fully reflect the equipment's operating status, and relying solely on a single parameter can easily lead to incomplete subsequent diagnosis. The characteristics of various parameters are independent; extracting features separately can accurately capture the core information of each parameter type, avoiding interference between different parameter types. Multiple features, being scattered and independent, cannot exert a synergistic effect. Fusing multiple features to obtain a comprehensive feature vector can integrate multi-source information, ensuring the accuracy and comprehensiveness of subsequent diagnosis.

[0018] For example, this embodiment collects electrical parameters through voltage and current sensors built into the electricity metering box. The voltage and current sensors are installed at the inlet and outlet terminals of the metering box, respectively, to collect voltage and current data in real time. Environmental parameters are collected through a built-in temperature and humidity sensor, installed in the middle of the metering box, to collect temperature and humidity data in real time. Box status parameters are collected through a dry contact sensor connected to the box door and box connectors to collect data such as door opening and closing and box vibration. Next, feature extraction is performed on the collected electrical parameters. Interference data is removed through filtering, and features such as the trend and fluctuation amplitude of voltage and current are extracted. Feature extraction is also performed on environmental parameters, including the rate of change and stable range of temperature and humidity. Feature extraction is performed on box status parameters, including the frequency of door opening and closing and vibration amplitude, resulting in multiple types of features. Finally, the obtained features are fused using a series splicing method, integrating the electrical parameter features, environmental parameter features, and box status parameter features in a preset order to form a comprehensive feature vector.

[0019] This embodiment achieves comprehensive collection of multi-dimensional information on equipment operation by acquiring electrical parameters, environmental status parameters, and cabinet status parameters of the power metering box, avoiding the limitations of single-parameter collection. Feature extraction of each type of parameter accurately captures core information, providing a reliable basis for subsequent diagnosis. Fusion of multiple features yields a comprehensive feature vector, enabling the synergistic utilization of multi-source information and improving data integrity and effectiveness.

[0020] In this embodiment, feature extraction is performed on electrical parameters, environmental state parameters, and enclosure state parameters to obtain multiple types of features, including: Electrical parameters are used to extract electrical time-domain features and electrical frequency-domain features; Environmental state features are obtained by extracting features from environmental state parameters; Feature extraction is performed on the state parameters of the enclosure to obtain the mechanical state features of the enclosure; Among them, electrical time domain characteristics, electrical frequency domain characteristics, environmental state characteristics, and enclosure mechanical state characteristics are all types of characteristics.

[0021] In this embodiment, electrical time-domain features are features extracted from electrical parameters that reflect the time-dimensional changes of these parameters, such as voltage fluctuation amplitude. Electrical frequency-domain features are features extracted from electrical parameters that reflect the frequency-dimensional changes of these parameters, such as current harmonic characteristics. Environmental state features are key information extracted from environmental state parameters, such as the rate of temperature change. Enclosure mechanical state features are key information extracted from enclosure state parameters, such as the opening and closing frequency of the enclosure door. The operating characteristics of electrical parameters have both time and frequency dimensions; extracting electrical time-domain and electrical frequency-domain features can comprehensively reflect the operating state of the electrical parameters.

[0022] For example, this embodiment extracts features from electrical parameters. Interference data is removed through filtering, and information such as the fluctuation amplitude and trend of voltage and current per unit time are extracted as electrical time-domain features. Information such as the harmonic distribution and frequency variation of voltage and current are extracted as electrical frequency-domain features. Next, features are extracted from environmental state parameters. By analyzing the trends of temperature and humidity data, information such as the rate of change and stable range of temperature and humidity are extracted as environmental state features. Finally, features are extracted from the enclosure state parameters. By statistically analyzing the number of door opening and closing times and vibration amplitude, information such as the door opening and closing frequency and vibration duration are extracted as mechanical state features of the enclosure, thus obtaining multiple types of features.

[0023] In this embodiment, multiple types of features are fused to obtain a comprehensive feature vector, including: The electrical time-domain characteristics are normalized to obtain standardized electrical time-domain characteristics. The electrical frequency domain characteristics are normalized to obtain standardized electrical frequency domain characteristics. The standardized electrical time-domain features and standardized electrical frequency-domain features are serially spliced ​​together to obtain integrated electrical features; Based on the correlation between integrated electrical characteristics and environmental state characteristics, the environmental-electrical coupling characteristics are obtained; The environmental-electrical coupling characteristics are denoised by filtering to obtain the denoised environmental-electrical coupling characteristics. The integrated electrical features and the denoised environment-electrical coupling features are weighted and fused dimension by dimension to obtain the electrical-environment synergistic features; The electrical-environmental co-operation characteristics and the mechanical state characteristics of the enclosure are concatenated and spliced ​​to obtain a comprehensive feature vector.

[0024] In this embodiment, the integrated electrical features and the denoised environment-electrical coupling features are weighted and fused dimension by dimension to obtain electrical-environment collaborative features, including: The integrated electrical features are analyzed to obtain the number of dimensions of the integrated electrical features and the feature values ​​corresponding to each dimension; The dimensionality of the denoised environmental-electrical coupling features is analyzed to obtain the number of dimensions of the denoised environmental-electrical coupling features and the feature values ​​corresponding to each dimension. Dimension matching parameters are obtained by performing dimension matching processing based on the number of dimensions of the integrated electrical features and the number of dimensions of the denoised environment-electrical coupling features. The denoised environment-electric coupling features are adapted to the dimension based on the dimension matching parameters to obtain the dimension-adapted environment-electric coupling features. Based on historical fault sample data, the fault contribution of each dimension in the integrated electrical features is analyzed to obtain the weight coefficients of each dimension of the integrated electrical features. Based on historical fault sample data, the fault contribution of each dimension in the environmental-electrical coupling characteristics after dimension adaptation is analyzed to obtain the weight coefficient of each dimension of the coupling characteristics. The feature values ​​corresponding to each dimension of the integrated electrical features and the feature values ​​corresponding to each dimension of the environment-electrical coupling features after dimension adaptation are multiplied by their respective weight coefficients and then summed to obtain multi-dimensional fused feature values. The feature values ​​of multiple dimensions are integrated in the order of their corresponding dimensions to obtain the electrical-environmental collaborative features.

[0025] In this embodiment, normalization is the process of processing features to bring them to the same order of magnitude, such as adjusting the values ​​of electrical time-domain features to a preset range. Standardized electrical time-domain features are features obtained after normalizing electrical time-domain features. Standardized electrical frequency-domain features are features obtained after normalizing electrical frequency-domain features. Serial splicing is the process of connecting and integrating two types of features in a preset order. Integrated electrical features are the set of electrical features obtained after serial splicing. Environment-electrical coupling features are features reflecting the relationship between electrical features and environmental state features. Denoising filtering is the process of removing interference information from features. Denoising-filtered environment-electrical coupling features are environment-electrical coupling features after denoising filtering. Dimensionally weighted fusion is the process of assigning weights to each dimension of the feature and then fusing them. Electrical-environmental collaborative features are features obtained after dimensionally weighted fusion. Dimensional parsing is the process of analyzing the number of feature dimensions and the feature values ​​of each dimension. The number of dimensions is the total number of dimensions contained in the feature. The feature values ​​are the specific numerical values ​​corresponding to each dimension. Dimension matching is the process of adapting the number of dimensions between two types of features. Dimension matching parameters are the parameters obtained from dimension matching. Dimension adaptation is the process of adjusting feature dimensions to fit another type of feature. The dimension-adapted environmental-electrical coupling features are the environmental-electrical coupling features after dimension adaptation. Historical fault sample data are the feature and fault information data related to past power metering box faults. Fault contribution analysis is the process of analyzing the degree of influence of each feature dimension on fault judgment. Weight coefficients are values ​​reflecting the importance of each feature dimension. The dimension fusion feature value is the value obtained by weighted summation of each dimension.

[0026] Considering the difference in magnitude between electrical time-domain and electrical frequency-domain characteristics, normalization can eliminate the magnitude effect and ensure the effectiveness of their splicing and fusion. Electrical and environmental factors influence each other; by correlating and integrating electrical features with environmental state features, we can obtain environmental-electrical coupling features, which can capture their synergistic effect.

[0027] For example, in this embodiment, the electrical time-domain features and electrical frequency-domain features are normalized separately. A linear scaling method is used to adjust the values ​​of both types of features to a preset range of 0 to 1, resulting in standardized electrical time-domain features and standardized electrical frequency-domain features. Next, the standardized electrical time-domain features and standardized electrical frequency-domain features are concatenated and spliced. Following a preset order of first standardizing the electrical time-domain features and then the electrical frequency-domain features, the dimensions of the two types of features are sequentially connected to obtain integrated electrical features. Then, based on the integrated electrical features and environmental state features, the correlation between electrical parameters and environmental parameters is captured by analyzing the changing trends of the two types of features, resulting in environmental-electrical coupling features. A low-pass filter is used to denoise the environmental-electrical coupling features, removing high-frequency interference information, resulting in denoised environmental-electrical coupling features. Next, dimensionality analysis was performed on the integrated electrical features and the denoised environmental-electrical coupling features, respectively. The number of dimensions for both types of features was counted, and the feature values ​​corresponding to each dimension were extracted. Based on the number of dimensions for both types of features, zero-padding was used for dimension matching to obtain dimension matching parameters. According to the dimension matching parameters, the denoised environmental-electrical coupling features were zero-padding and adapted to obtain dimension-adapted environmental-electrical coupling features. Then, based on historical fault sample data, the correlation between each dimension feature and the fault was statistically analyzed. Fault contribution analysis was performed on each dimension of the integrated electrical features and the dimension-adapted environmental-electrical coupling features to obtain their respective weight coefficients. Finally, the feature values ​​of each dimension of the integrated electrical features and the corresponding weight coefficients were multiplied sequentially, and the feature values ​​of each dimension of the dimension-adapted environmental-electrical coupling features and the corresponding weight coefficients were multiplied sequentially. The sum of each set of products yielded multiple dimensional fusion feature values. These feature values ​​were integrated in the corresponding dimensional order to obtain the electrical-environmental collaborative features. The electrical-environmental collaborative features and the mechanical state features of the enclosure were concatenated and spliced ​​to obtain a comprehensive feature vector.

[0028] This embodiment eliminates numerical magnitude differences by normalizing electrical time-domain and frequency-domain features, ensuring the rationality of splicing and fusion. Series splicing yields integrated electrical features, achieving unification of electrical features. Correlation yields environment-electrical coupling features, capturing the synergistic effect between electrical and environmental factors; denoising and filtering improve feature purity. Dimensional matching and adaptation ensure smooth dimensional weighted fusion, and weight coefficients based on historical samples make the fusion more targeted. The comprehensive feature vector obtained through multi-step fusion integrates the core information of various features, improving data completeness and effectiveness, providing a precise and reliable foundation for subsequent fault diagnosis, and effectively improving the accuracy and comprehensiveness of diagnosis.

[0029] S102: Compare each feature component in each type of feature with its corresponding preset benchmark feature, filter out the feature components that are abnormal, and construct a suspected fault feature vector based on the feature components that are abnormal.

[0030] In this embodiment, each feature component in various features is compared with its corresponding preset benchmark feature to filter out feature components that are abnormal, including: Recall the preset benchmark features corresponding to each feature component from the preset benchmark feature library; The difference between each feature component and its corresponding preset benchmark feature is calculated and the absolute value is converted to obtain the absolute value of the difference between each feature component. Determine the anomaly threshold corresponding to each preset benchmark feature to obtain the anomaly threshold for each feature; The absolute value of each feature difference is compared and marked with the corresponding feature anomaly threshold in turn to obtain multiple feature comparison results; the multiple feature comparison results are then filtered to obtain the feature components that are abnormal.

[0031] In this embodiment, the various features are electrical time-domain features, electrical frequency-domain features, environmental state features, and enclosure mechanical state features. Feature components are the specific dimensional information contained in each feature, such as the voltage fluctuation amplitude dimension in the electrical time-domain features. Preset benchmark features are features set in advance and used as comparison standards, such as the voltage fluctuation amplitude standard under normal operating conditions. Comparison is the process of comparing and analyzing feature components with preset benchmark features. Filtering is the process of selecting components that meet the abnormal conditions from all feature components. An abnormal feature component is a feature component whose deviation from the preset benchmark feature exceeds the allowable range. The suspected fault feature vector is a data format formed based on abnormal feature components and used for subsequent fault judgment. The preset benchmark feature library is a database storing various preset benchmark features, such as a database storing various feature benchmark values ​​under different operating conditions. Difference calculation is the process of calculating the numerical difference between a feature component and a preset benchmark feature. Absolute value conversion is the process of converting the difference into a non-negative value. The absolute value of the feature difference is the value obtained after difference calculation and absolute value conversion. The abnormal threshold is a pre-set critical value for judging whether a feature component is abnormal. The feature anomaly threshold is the critical value for anomaly judgment corresponding to each preset benchmark feature. Labeling is the process of classifying and identifying the feature comparison results. The feature comparison result is the judgment result obtained by comparing the absolute value of the feature difference with the feature anomaly threshold.

[0032] Considering that various features need to be compared with standards to determine whether they are abnormal, setting preset benchmark features and establishing a preset benchmark feature library can achieve unified access to comparison standards and ensure consistency in comparisons. Difference alone cannot accurately reflect the degree of deviation; absolute value conversion can eliminate the influence of positive and negative deviations, ensuring the rationality of deviation judgment. Different feature components have different anomaly judgment criteria; setting corresponding feature anomaly thresholds for each preset benchmark feature can improve the targeting and accuracy of anomaly screening. This embodiment, through comparison, marking, and screening steps, can systematically filter out feature components with anomalies, construct suspected fault feature vectors based on them, and provide accurate anomaly data support for subsequent fault probability calculation and secondary judgment, avoiding interference from invalid features in the diagnostic results.

[0033] For example, this embodiment calls a preset benchmark feature library, which stores standard feature values ​​corresponding to various feature components. Based on the type of each feature component to be compared, the corresponding preset benchmark feature is retrieved from the library to ensure that each feature component has a matching comparison standard. Next, the difference between each feature component and its corresponding preset benchmark feature is calculated. The specific value of each feature component is subtracted from the value of the corresponding preset benchmark feature, and the calculated difference is then converted to an absolute value, converting negative deviations to non-negative values, resulting in the absolute value of each feature difference. Then, the feature anomaly threshold corresponding to each preset benchmark feature is determined. The feature anomaly threshold is set based on historical data of the normal operation of the electricity metering box. Different types of feature components correspond to different feature anomaly thresholds; for example, the anomaly threshold for electrical time-domain features is higher than that for environmental state features. Next, the absolute value of each feature difference is compared sequentially with the corresponding feature anomaly threshold. If the absolute value of the feature difference is greater than the corresponding feature anomaly threshold, it is marked as abnormal; otherwise, it is marked as normal, resulting in multiple feature comparison results. Finally, the results of multiple feature comparisons are filtered to select the feature components corresponding to the feature comparison results marked as abnormal. These are the feature components with abnormalities. All the feature components with abnormalities are then integrated in a preset order to construct a suspected fault feature vector.

[0034] This embodiment ensures the uniformity and accuracy of comparison standards by calling corresponding preset benchmark features from a preset benchmark feature library, avoiding misjudgments caused by inconsistent comparison standards. Difference calculation and absolute value conversion accurately reflect the degree of deviation between the feature components and the benchmark, eliminating the influence of positive and negative deviations. Setting corresponding feature anomaly thresholds for each preset benchmark feature improves the targeting of anomaly screening. Through the systematic steps of comparison, marking, and screening, abnormal feature components can be accurately screened out. Based on these, a suspected fault feature vector is constructed, providing accurate anomaly data support for subsequent fault diagnosis, effectively reducing false positives and false negatives, and improving the accuracy and efficiency of subsequent fault diagnosis.

[0035] S103: Determine the weight of each suspected fault feature component in the suspected fault feature vector, and determine the fault probability of each suspected fault feature component based on the weight.

[0036] In this embodiment, determining the weight of each suspected fault feature component in the suspected fault feature vector includes: Each suspected fault feature component is identified by its feature type to obtain the feature type corresponding to each suspected fault feature component. For each suspected fault feature component, based on the feature type corresponding to the suspected fault feature component, retrieve the fault correlation degree corresponding to the feature type from the preset feature type-fault association library; The fault correlation degree of each suspected fault feature component is normalized to obtain the corresponding normalized fault correlation degree. The degree of abnormality of each suspected fault feature component is quantified to obtain its corresponding abnormal quantification value; the abnormal quantification value is then normalized to obtain its corresponding standardized abnormal quantification value. The weights of each suspected fault feature component are obtained by weighting each normalized fault correlation degree and its corresponding standardized anomaly quantification value.

[0037] In this embodiment, the suspected fault feature component is a specific abnormal feature dimension contained in the suspected fault feature vector, such as an abnormal voltage fluctuation amplitude component. The weight is a numerical value reflecting the importance of each suspected fault feature component; for example, components with a high correlation to the fault have a larger weight. The fault probability is the likelihood of the fault corresponding to each suspected fault feature component occurring, determined based on its weight. Feature type identification is the process of determining the feature category to which each suspected fault feature component belongs. Feature type refers to the feature category to which the suspected fault feature component belongs, such as an electrical time-domain feature type. The preset feature type-fault association library is a database storing the correlation degree between various feature types and faults. Fault correlation degree is the degree of close association between a feature type and a fault. Normalization is the process of adjusting the fault correlation degree to the same order of magnitude. Normalized fault correlation degree is the fault correlation degree after normalization. Anomaly degree quantification is the process of converting the anomaly degree of the suspected fault feature component into a specific numerical value. The anomaly quantization value is the anomaly degree value obtained after quantization. The standardized anomaly quantization value is the value obtained after normalizing the anomaly quantization value.

[0038] Considering the varying degrees of influence of different suspected fault feature components on fault judgment, determining the weight of each component makes subsequent weighted calculations more targeted and avoids judgment bias caused by treating all components equally. The feature type determines its correlation with the fault; retrieving the fault correlation degree through feature type identification ensures the rationality of weight allocation. Fault correlation degree and anomaly quantification value have different magnitudes; normalization can eliminate magnitude interference and ensure the effectiveness of weighted calculations. Weights need to comprehensively reflect both the degree of correlation and the degree of anomaly; weighting these two factors makes the weights more closely match the actual fault judgment needs, resulting in a more accurate fault probability and providing a reliable basis for subsequent fault investigation and secondary judgment.

[0039] For example, this embodiment identifies the feature type of each suspected fault feature component in the suspected fault feature vector. By analyzing the source and attributes of each component, its corresponding feature type is determined. For example, a component originating from electrical time-domain features corresponds to an electrical time-domain feature type. Next, a preset feature type-fault association library is invoked. This database is built based on past fault data and stores the correlation degree between various feature types and different faults. Based on the feature type corresponding to each suspected fault feature component, the corresponding fault correlation degree is retrieved. Then, the fault correlation degree of each suspected fault feature component is normalized by using linear scaling to adjust the correlation degree value to the range of 0 to 1, obtaining the corresponding normalized fault correlation degree. Next, the anomaly degree of each suspected fault feature component is quantified. By comparing the deviation of the component from a preset benchmark feature, the anomaly degree is converted into a specific value, and then the anomaly quantification value is normalized to obtain a standardized anomaly quantification value. Afterward, the normalized fault correlation degree and the corresponding standardized anomaly quantification value are weighted and calculated. The weight ratios of the two are preset, and the two values ​​are multiplied by their respective ratios and then summed to obtain the weight of each suspected fault feature component. Finally, the failure probability of each suspected fault feature component is determined based on this weight.

[0040] This embodiment can identify and retrieve fault correlation through feature type, ensuring that the weight allocation aligns with fault diagnosis needs and improving the rationality of the weights. Normalization eliminates the magnitude difference between fault correlation and anomaly quantification values, guaranteeing the accuracy of weighted calculations. Determining weights by comprehensively considering both correlation and anomaly levels makes the weights more targeted and scientific. The fault probability obtained based on these weights accurately reflects the fault likelihood corresponding to each suspected fault feature component, providing reliable support for subsequent fault comparison and secondary judgment, effectively improving the accuracy of fault diagnosis and reducing misjudgments and omissions.

[0041] In this embodiment, the formula for calculating the fault probability of each suspected fault feature component based on this weight is as follows:

[0042] in, For the first The fault probability corresponding to each suspected fault feature component is in the range of [0,1]. For the first The weights of each suspected fault feature component; For the first Standardized anomaly quantification values ​​of one suspected fault feature component; This is a dynamic correction coefficient with a value range of [0.05, 0.15]. It can be dynamically adjusted according to the fault correlation degree; the higher the fault correlation degree, the larger the coefficient. For the first Normalized fault correlation degree of each suspected fault feature component.

[0043] Weighted terms are This product term directly incorporates the weights of the suspected fault feature components. Standardized outlier quantification value The secondary weighting reflects the core impact of weight and anomaly level on failure probability; that is, the higher the weight and the greater the anomaly level, the higher the failure probability. The higher; the correction term is Normalized fault correlation degree was introduced. With dynamic correction coefficient The calculation results of the weighted terms are further revised so that the failure probability depends not only on the weight and degree of anomaly of the feature, but also on the essential correlation between the feature type and the failure. This effectively avoids misjudgment caused by relying solely on the degree of anomaly and improves the accuracy and rationality of the failure probability calculation.

[0044] S104: Compare the fault probability of each suspected fault feature component with its corresponding preset probability threshold. For suspected fault feature components whose fault probability is less than the corresponding preset probability threshold, extract the associated features corresponding to the suspected fault feature component from the comprehensive feature vector, and determine whether the power metering box is faulty based on the associated features.

[0045] In this embodiment, the preset probability threshold is a pre-set critical value used to determine whether the fault probability requires secondary verification, for example, 0.7. The comparison is the process of numerically comparing the fault probability with the preset probability threshold. The associated features are other feature components in the comprehensive feature vector that are correlated with the suspected fault feature component, such as temperature feature components related to voltage anomalies. The electricity metering box is a cabinet device used for electricity metering. A fault is an abnormal state that occurs during the operation of the electricity metering box, such as inaccurate metering or overheating of the box.

[0046] Considering that the probability assessment of a single fault may be affected by noise or random factors, posing a risk of misjudgment, a preset probability threshold is set as the judgment benchmark to distinguish between high-confidence and low-confidence suspected faults. The judgment of low-confidence suspected faults lacks sufficient basis and requires the introduction of more related information for supplementary verification. Related features are extracted from the comprehensive feature vector, and cross-validation using multi-dimensional information can be utilized. Directly excluding low-probability suspected faults may lead to missed diagnoses. Secondary judgment through related features can improve the accuracy of diagnostic results without increasing the complexity of the model, balancing the risks of misjudgment and missed diagnoses.

[0047] For example, this embodiment calls a pre-set probability threshold corresponding to each suspected fault feature component. The preset probability threshold is set based on historical fault data and diagnostic experience, and different thresholds correspond to different types of suspected fault feature components. Next, the fault probability of each suspected fault feature component is compared with the corresponding preset probability threshold to determine if the fault probability is less than the threshold. When the fault probability is less than the corresponding preset probability threshold, the associated features corresponding to the suspected fault feature component are extracted from the comprehensive feature vector. The extraction of associated features is based on preset feature association rules; for example, the associated features corresponding to a voltage anomaly component include current features and box temperature features. Then, the extracted associated features are analyzed to observe the changing trends and numerical states of the associated features and determine if there are any abnormal changes in the associated features. Finally, based on the analysis results of the associated features, a comprehensive judgment is made as to whether the electricity metering box has the suspected fault. If the associated features also show abnormal changes, the electricity metering box is determined to have a fault; if the associated features show no obvious abnormalities, the suspected fault is ruled out.

[0048] This embodiment can quickly distinguish between high-confidence and low-confidence suspected faults by comparing the fault probability with a preset probability threshold, reducing unnecessary secondary verification. For low-confidence suspected faults, it extracts associated features for secondary judgment, utilizing multi-dimensional information cross-validation to effectively eliminate false alarms caused by noise or random factors. Simultaneously, it avoids missed detections due to directly excluding low-probability suspected faults, improving the accuracy and reliability of fault diagnosis and ensuring the stable operation of the electricity metering box.

[0049] As can be seen from the above, this embodiment of the application simultaneously acquires the electrical parameters, environmental status parameters, and box status parameters of the power metering box. Features are extracted from these parameters and then fused to form a comprehensive feature vector, achieving comprehensive integration of multi-source information. This avoids the limitations of single-parameter diagnosis and improves the comprehensiveness of diagnosis from the source. This embodiment of the application compares various features with preset benchmarks to screen for anomalies, constructs suspected fault feature vectors, and assigns weights based on feature correlation and anomaly severity to calculate fault probability, achieving quantitative fault judgment, which is more accurate than existing qualitative judgment methods. For low-probability suspected faults, related features are extracted from the comprehensive feature vector for secondary verification, effectively eliminating false positives, further improving diagnostic accuracy, and reducing ineffective maintenance and fault expansion caused by false positives and missed faults.

[0050] In one embodiment of this application, the intelligent fault diagnosis method for electricity metering boxes, after determining whether the electricity metering box is faulty based on correlation features, further includes: When the electricity metering box malfunctions, the preset fault feature-fault type mapping library is queried to obtain the specific fault type corresponding to the fault. The data acquisition node information corresponding to the fault is parsed to obtain the specific location information of the fault occurrence; By integrating specific fault types and location information, a fault diagnosis report is obtained.

[0051] In this embodiment, the fault refers to an abnormal state that occurs during the operation of the electricity metering box, such as inaccurate metering or overheating of the box. The preset fault feature-fault type mapping library is a pre-established database that stores the correspondence between fault features and specific fault types, such as storing the correspondence between voltage anomaly features and metering error faults. Specific fault types refer to the specific categories of faults, such as metering error faults or box sealing faults. Acquisition node information refers to the identifiers and location information corresponding to the nodes used to collect various parameters, such as the installation node information of voltage sensors. Parsing is the process of analyzing and processing the acquired node information to extract key information. Specific location information refers to the specific location where the fault occurs, such as the installation location of the voltage sensor inside the metering box. Integration is the process of organizing and merging the specific fault type and specific location information. The fault diagnosis report is a formal document containing key fault information, used to present the fault diagnosis results.

[0052] Considering that simply identifying a fault in the electricity metering box is insufficient for maintenance needs, maintenance personnel must clearly define the specific fault type to implement targeted measures. Therefore, a pre-defined fault characteristic-fault type mapping library is established to enable rapid matching of fault characteristics with specific fault types. Fault location directly impacts maintenance efficiency; parsing the collected node information allows for precise fault location pinpointing, preventing maintenance personnel from blindly troubleshooting. Maintenance work requires standardized diagnostic records. Integrating specific fault types and location information generates fault diagnosis reports that clearly present diagnostic results, supporting subsequent maintenance, review, and data retention, ensuring a complete closed loop in fault diagnosis.

[0053] For example, when a fault is determined in the electricity metering box based on associated features, a preset fault feature-fault type mapping library is invoked. This mapping library is built based on historical fault sample data and stores the correspondence between various fault features and specific fault types. By matching the currently existing fault features with the features in the mapping library, the specific fault type corresponding to the fault is retrieved. Next, the data acquisition node information corresponding to the fault is retrieved. This data acquisition node information is pre-stored in the data acquisition system and includes the identifier, installation location, and acquisition parameter type of each data acquisition node. The data acquisition node information is parsed to extract key information such as the installation orientation and module to which the data acquisition node belongs, thus obtaining the specific location information of the fault. Then, the obtained specific fault type and specific location information are integrated, clearly marking the fault type name, a detailed description of the fault location, and supplementing the corresponding feature information and judgment basis. Finally, according to the preset report format, the integrated information is filled into the corresponding module to generate a fault diagnosis report. The report clearly presents the specific fault type, specific location information, and diagnostic basis, completing the entire fault diagnosis process.

[0054] This embodiment can quickly determine the specific type of fault by querying a preset fault feature-fault type mapping library, providing maintenance personnel with a clear direction for handling and avoiding blind maintenance. Parsing the collected node information can accurately locate the specific fault location, shortening fault investigation time and improving maintenance efficiency. Integrating specific fault type and location information to generate a fault diagnosis report can standardize the presentation of diagnostic results, providing a reliable basis for subsequent maintenance, review, and data retention, ensuring a complete closed loop in fault diagnosis, further improving the practicality and standardization of fault diagnosis, and ensuring the efficient restoration of normal operation of the power metering box.

[0055] Corresponding to the intelligent fault diagnosis method for the electricity metering box in the above embodiment, Figure 2 This is a structural block diagram of an intelligent fault diagnosis device for an electricity metering box provided in one embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2The intelligent fault diagnosis device 20 for the electricity metering box includes: a feature extraction module 21, a suspected fault determination module 22, a fault probability determination module 23, and a fault diagnosis module 24.

[0056] The feature extraction module 21 is used to obtain the electrical parameters, environmental status parameters and box status parameters of the power metering box; to extract features from the electrical parameters, environmental status parameters and box status parameters respectively to obtain multiple types of features, and to fuse the multiple types of features to obtain a comprehensive feature vector; The suspected fault determination module 22 is used to compare each feature component in various features with its corresponding preset benchmark features, filter out the feature components that are abnormal, and construct a suspected fault feature vector based on the feature components that are abnormal. The fault probability determination module 23 is used to determine the weight of each suspected fault feature component in the suspected fault feature vector, and to determine the fault probability of each suspected fault feature component based on the weight. The fault diagnosis module 24 is used to compare the fault probability of each suspected fault feature component with its corresponding preset probability threshold. For suspected fault feature components whose fault probability is less than the corresponding preset probability threshold, the module extracts the associated features corresponding to the suspected fault feature component from the comprehensive feature vector and determines whether the power metering box is faulty based on the associated features.

[0057] In one embodiment of this application, the feature extraction module 21 is specifically used for: Electrical parameters are used to extract electrical time-domain features and electrical frequency-domain features; Environmental state features are obtained by extracting features from environmental state parameters; Feature extraction is performed on the state parameters of the enclosure to obtain the mechanical state features of the enclosure; Among them, electrical time domain characteristics, electrical frequency domain characteristics, environmental state characteristics, and enclosure mechanical state characteristics are all types of characteristics.

[0058] In one embodiment of this application, the feature extraction module 21 is specifically used for: The electrical time-domain characteristics are normalized to obtain standardized electrical time-domain characteristics. The electrical frequency domain characteristics are normalized to obtain standardized electrical frequency domain characteristics. The standardized electrical time-domain features and standardized electrical frequency-domain features are serially spliced ​​together to obtain integrated electrical features; Based on the correlation between integrated electrical characteristics and environmental state characteristics, the environmental-electrical coupling characteristics are obtained; The environmental-electrical coupling characteristics are denoised by filtering to obtain the denoised environmental-electrical coupling characteristics. The integrated electrical features and the denoised environment-electrical coupling features are weighted and fused dimension by dimension to obtain the electrical-environment synergistic features; The electrical-environmental co-operation characteristics and the mechanical state characteristics of the enclosure are concatenated and spliced ​​to obtain a comprehensive feature vector.

[0059] In one embodiment of this application, the feature extraction module 21 is specifically used for: The integrated electrical features are analyzed to obtain the number of dimensions of the integrated electrical features and the feature values ​​corresponding to each dimension; The dimensionality of the denoised environmental-electrical coupling features is analyzed to obtain the number of dimensions of the denoised environmental-electrical coupling features and the feature values ​​corresponding to each dimension. Dimension matching parameters are obtained by performing dimension matching processing based on the number of dimensions of the integrated electrical features and the number of dimensions of the denoised environment-electrical coupling features. The denoised environment-electric coupling features are adapted to the dimension based on the dimension matching parameters to obtain the dimension-adapted environment-electric coupling features. Based on historical fault sample data, the fault contribution of each dimension in the integrated electrical features is analyzed to obtain the weight coefficients of each dimension of the integrated electrical features. Based on historical fault sample data, the fault contribution of each dimension in the environmental-electrical coupling characteristics after dimension adaptation is analyzed to obtain the weight coefficient of each dimension of the coupling characteristics. The feature values ​​corresponding to each dimension of the integrated electrical features and the feature values ​​corresponding to each dimension of the environment-electrical coupling features after dimension adaptation are multiplied by their respective weight coefficients and then summed to obtain multi-dimensional fused feature values. The feature values ​​of multiple dimensions are integrated in the order of their corresponding dimensions to obtain the electrical-environmental collaborative features.

[0060] In one embodiment of this application, the suspected fault determination module 22 is specifically used for: Recall the preset benchmark features corresponding to each feature component from the preset benchmark feature library; The difference between each feature component and its corresponding preset benchmark feature is calculated and the absolute value is converted to obtain the absolute value of the difference between each feature component. Determine the anomaly threshold corresponding to each preset benchmark feature to obtain the anomaly threshold for each feature; The absolute value of each feature difference is compared and marked with the corresponding feature anomaly threshold in turn to obtain multiple feature comparison results; the multiple feature comparison results are then filtered to obtain the feature components that are abnormal.

[0061] In one embodiment of this application, the fault probability determination module 23 is specifically used to: identify the feature type of each suspected fault feature component to obtain the feature type corresponding to each suspected fault feature component. For each suspected fault feature component, based on the feature type corresponding to the suspected fault feature component, retrieve the fault correlation degree corresponding to the feature type from the preset feature type-fault association library; The fault correlation degree of each suspected fault feature component is normalized to obtain the corresponding normalized fault correlation degree. The degree of abnormality of each suspected fault feature component is quantified to obtain its corresponding abnormal quantification value; the abnormal quantification value is then normalized to obtain its corresponding standardized abnormal quantification value. The weights of each suspected fault feature component are obtained by weighting each normalized fault correlation degree and its corresponding standardized anomaly quantification value.

[0062] In one embodiment of this application, the intelligent fault diagnosis device 20 for the electricity metering box further includes: The fault diagnosis report generation module is used to query the preset fault feature-fault type mapping library to obtain the specific fault type corresponding to the fault when there is a fault in the electricity metering box. The data acquisition node information corresponding to the fault is parsed to obtain the specific location information of the fault occurrence; By integrating specific fault types and location information, a fault diagnosis report is obtained.

[0063] It should be noted that the specific limitations of the above-described intelligent fault diagnosis device 20 for electricity metering boxes can be found in the limitations of the intelligent fault diagnosis method for electricity metering boxes described above, and will not be repeated here. Each module of the above device can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in the processor of the electronic device in hardware form or independent of the processor, or it can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.

[0064] This application provides an electronic device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps of the method provided in any optional embodiment of this application.

[0065] In one alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3The illustrated electronic device 300 includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may further include a transceiver 304, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 304 is not limited to one type, and the structure of this electronic device 300 does not constitute a limitation on the embodiments of this application.

[0066] Processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0067] Bus 302 may include a pathway for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0068] The memory 303 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium capable of carrying or storing computer programs and capable of being read by a computer, without limitation herein.

[0069] The memory 303 is used to store computer programs that execute the embodiments of this application, and the execution is controlled by the processor 301. The processor 301 is used to execute the computer programs stored in the memory 303 to implement the steps shown in the foregoing method embodiments.

[0070] In an exemplary embodiment, a computer-readable storage medium is also provided, which stores at least one computer program, which is loaded and executed by a processor of a computer device to enable the computer to implement any of the above-described intelligent fault diagnosis methods for electricity metering boxes.

[0071] In one possible implementation, the aforementioned computer-readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a solid-state drive (SSD), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc. The random access memory can include resistive random access memory (ReRAM) and dynamic random access memory (DRAM).

[0072] In an exemplary embodiment, a computer program or computer program product is also provided, which includes computer instructions loaded and executed by a processor to enable the computer to implement any of the above-described intelligent fault diagnosis methods for electricity metering boxes.

[0073] It should be noted that all information (including but not limited to user equipment information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in this application have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the electrical parameters, environmental condition parameters, and enclosure condition parameters involved in this application were all obtained with full authorization.

[0074] In other words, the data collection and processing in this application should strictly comply with the requirements of relevant national laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0075] It should be further noted that the terms "first," "second," etc., used in this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. The implementation methods described in the above exemplary embodiments do not represent all implementation methods consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application.

[0076] It should be understood that "multiple" as used in this article refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0077] Furthermore, the step numbers described herein are merely illustrative of one possible execution order between steps. In some other embodiments, the steps may not be executed in the order of their numbers, such as two steps with different numbers being executed simultaneously, or two steps with different numbers being executed in the reverse order of the illustration. This application does not limit this.

[0078] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. Optionally, the program is stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0079] The above description is merely an exemplary embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A method for intelligent fault diagnosis of an electricity metering box, characterized in that, include: Obtain the electrical parameters, environmental condition parameters, and cabinet condition parameters of the electricity metering box; Feature extraction is performed on the electrical parameters, environmental state parameters, and enclosure state parameters respectively to obtain multiple types of features, and the multiple types of features are fused to obtain a comprehensive feature vector; Each feature component in various features is compared with its corresponding preset benchmark feature to filter out the feature components that are abnormal, and a suspected fault feature vector is constructed based on the abnormal feature components. Determine the weight of each suspected fault feature component in the suspected fault feature vector, and determine the fault probability of each suspected fault feature component based on the weight. The failure probability of each suspected fault feature component is compared with its corresponding preset probability threshold. For suspected fault feature components whose failure probability is less than the corresponding preset probability threshold, the associated features corresponding to the suspected fault feature component are extracted from the comprehensive feature vector. Based on the associated features, it is determined whether the power metering box is faulty.

2. The intelligent fault diagnosis method for electricity metering boxes as described in claim 1, characterized in that, The electrical parameters, environmental state parameters, and enclosure state parameters are each subjected to feature extraction to obtain multiple types of features, including: The electrical parameters are subjected to feature extraction to obtain electrical time-domain features and electrical frequency-domain features; Environmental state features are obtained by extracting features from the environmental state parameters; Feature extraction is performed on the state parameters of the enclosure to obtain the mechanical state features of the enclosure; Among them, the electrical time domain characteristics, the electrical frequency domain characteristics, the environmental state characteristics, and the enclosure mechanical state characteristics are all one type of multiple characteristics.

3. The intelligent fault diagnosis method for electricity metering boxes as described in claim 2, characterized in that, The process of fusing multiple types of features to obtain a comprehensive feature vector includes: The electrical time-domain features are normalized to obtain standardized electrical time-domain features; The electrical frequency domain features are normalized to obtain standardized electrical frequency domain features; The standardized electrical time-domain features and the standardized electrical frequency-domain features are concatenated and spliced ​​together to obtain integrated electrical features; Based on the correlation between the integrated electrical characteristics and the environmental state characteristics, the environmental-electrical coupling characteristics are obtained; The environmental-electrical coupling characteristics are subjected to denoising filtering to obtain denoised environmental-electrical coupling characteristics; The integrated electrical features and the denoised environment-electrical coupling features are weighted and fused dimension by dimension to obtain the electrical-environment collaborative features; The electrical-environmental coordination features and the mechanical state features of the enclosure are concatenated and spliced ​​together to obtain the comprehensive feature vector.

4. The intelligent fault diagnosis method for electricity metering boxes as described in claim 3, characterized in that, The step of performing a dimension-by-dimensional weighted fusion of the integrated electrical features and the denoised environment-electrical coupling features to obtain electrical-environment collaborative features includes: The integrated electrical features are analyzed to obtain the number of dimensions of the integrated electrical features and the feature values ​​corresponding to each dimension; The denoised environmental-electrical coupling features are analyzed for dimensionality to obtain the number of dimensions of the denoised environmental-electrical coupling features and the feature values ​​corresponding to each dimension. Dimension matching is performed based on the number of dimensions of the integrated electrical features and the number of dimensions of the denoised environment-electrical coupling features to obtain dimension matching parameters. The denoised environment-electric coupling features are subjected to dimension adaptation processing based on the dimension matching parameters to obtain dimension-adapted environment-electric coupling features. Based on historical fault sample data, the fault contribution of each dimension in the integrated electrical features is analyzed to obtain the weight coefficients of each dimension of the integrated electrical features. Based on historical fault sample data, the fault contribution of each dimension in the environmental-electrical coupling characteristics after dimension adaptation is analyzed to obtain the weight coefficient of each dimension of the coupling characteristics. The feature values ​​corresponding to each dimension of the integrated electrical features and the feature values ​​corresponding to each dimension of the environment-electrical coupling features after dimension adaptation are multiplied by their respective weight coefficients and then summed to obtain multiple dimension fused feature values. The electrical-environmental collaborative features are obtained by integrating the feature values ​​of multiple dimensions in the order of their corresponding dimensions.

5. The intelligent fault diagnosis method for electricity metering boxes as described in claim 1, characterized in that, The step of comparing each feature component in various features with its corresponding preset benchmark feature to filter out feature components that are abnormal includes: Recall the preset benchmark features corresponding to each feature component from the preset benchmark feature library; The difference between each feature component and its corresponding preset benchmark feature is calculated and the absolute value is converted to obtain the absolute value of the difference between each feature component. Determine the anomaly threshold corresponding to each of the preset benchmark features to obtain the anomaly threshold for each feature; The absolute value of each feature difference is compared and marked with the corresponding feature anomaly threshold in turn to obtain multiple feature comparison results; the multiple feature comparison results are then filtered to obtain the feature components that have anomalies.

6. The intelligent fault diagnosis method for electricity metering boxes as described in claim 1, characterized in that, Determining the weights of each suspected fault feature component in the suspected fault feature vector includes: Each suspected fault feature component is identified by its feature type to obtain the feature type corresponding to each suspected fault feature component. For each suspected fault feature component, based on the feature type corresponding to the suspected fault feature component, retrieve the fault correlation degree corresponding to the feature type from the preset feature type-fault association library; The fault correlation degree of each suspected fault feature component is normalized to obtain the corresponding normalized fault correlation degree. Each suspected fault feature component is quantified to obtain its corresponding abnormal quantification value; each abnormal quantification value is then normalized to obtain its corresponding standardized abnormal quantification value. The weights of each suspected fault feature component are obtained by weighting each normalized fault correlation degree and its corresponding standardized anomaly quantification value.

7. The intelligent fault diagnosis method for electricity metering boxes as described in claim 1, characterized in that, After determining whether the electricity metering box is faulty based on the associated features, the method further includes: When the electricity metering box is faulty, the preset fault feature-fault type mapping library is queried to obtain the specific fault type corresponding to the fault. The data acquisition node information corresponding to the fault is parsed to obtain the specific location information of the fault occurrence; The specific fault type and the specific location information are integrated to obtain a fault diagnosis report.

8. A fault intelligent diagnostic device for an electricity metering box, characterized in that, include: The feature extraction module is used to acquire the electrical parameters, environmental status parameters, and box status parameters of the power metering box; to extract features from the electrical parameters, environmental status parameters, and box status parameters respectively to obtain multiple types of features, and to fuse the multiple types of features to obtain a comprehensive feature vector; The suspected fault determination module is used to compare each feature component in various features with its corresponding preset benchmark features, filter out the feature components that are abnormal, and construct a suspected fault feature vector based on the abnormal feature components. The fault probability determination module is used to determine the weight of each suspected fault feature component in the suspected fault feature vector, and to determine the fault probability of each suspected fault feature component based on the weight. The fault diagnosis module is used to compare the fault probability of each suspected fault feature component with its corresponding preset probability threshold. For suspected fault feature components whose fault probability is less than the corresponding preset probability threshold, the module extracts the associated features corresponding to the suspected fault feature component from the comprehensive feature vector and determines whether the power metering box is faulty based on the associated features.

9. An electronic device, characterized in that, The electronic device includes a processor and a memory, the memory storing a computer program, which is loaded and executed by the processor to implement the intelligent fault diagnosis method for the power metering box as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which is loaded and executed by a processor to implement the intelligent fault diagnosis method for electricity metering boxes as described in any one of claims 1 to 7.

Images