Method and device for detecting abnormality of archive of low-voltage customer, electronic equipment and medium

Abstract

Embodiments of the present application provide a kind of low-voltage customer's archive abnormality detection method, device, electronic equipment and medium.The method comprises: obtaining the archive database of low-voltage customer;Archive database includes at least one archive data;Each archive data in archive database is carried out abnormality detection processing, and abnormality result is obtained;And abnormality result is marked with each archive data, and archive marked data is obtained;Based on interval two type fuzzy comprehensive evaluation model, quality evaluation processing is carried out to archive marked data, and archive grade data is obtained;Wherein, archive grade data represents the quality grade of each archive data in archive database;Based on archive marked data and archive grade data, generate archive abnormality report;Wherein, archive abnormality report represents archive database exception positioning and risk characteristic analysis report.The method is used to achieve the effect of efficiently and accurately detecting the archive of low-voltage customer.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, electronic device and medium for detecting file anomalies in low-voltage customers. Background Technology

[0002] In the field of low-voltage customer file management in the power industry, the completeness and accuracy of file information are directly related to the accuracy of electricity bill calculation and marketing security.

[0003] In related technologies, anomaly detection in low-voltage customer files mainly relies on manual verification. However, manual verification suffers from inefficiency, high subjectivity, and difficulty in ensuring consistency when dealing with massive amounts of file data.

[0004] Therefore, there is an urgent need for an efficient and accurate solution for detecting anomalies in the files of low-voltage customers. Summary of the Invention

[0005] This application provides a method, apparatus, electronic device, and medium for detecting anomalies in the files of low-voltage customers, so as to achieve efficient and accurate anomaly detection in the files of low-voltage customers.

[0006] In a first aspect, embodiments of this application provide a method for detecting abnormal records of low-voltage customers, including:

[0007] Obtain a database of low-voltage customer records; the database includes at least one record.

[0008] Anomaly detection processing is performed on each piece of archival data in the archival database to obtain anomaly results; and the anomaly results are then marked with each piece of archival data to obtain archival marked data; wherein, the archival marked data represents the archival data associated with the anomaly results;

[0009] Based on the interval type II fuzzy comprehensive evaluation model, the archival tag data is subjected to quality assessment processing to obtain archival grade data; wherein, the archival grade data represents the quality grade of each archival data in the archival database;

[0010] Based on the archive tagging data and the archive level data, an archive anomaly report is generated; wherein, the archive anomaly report represents the anomaly location and risk characteristic analysis report of the archive database.

[0011] In one possible implementation, the archive data includes archive tables, work order tables, and settlement tables; anomaly detection processing is performed on each piece of archive data in the archive database to obtain anomaly results, including:

[0012] Feature extraction processing is performed on the archive table, work order table, and settlement table of each archive data in the archive database to obtain an associated feature vector; wherein, the associated feature vector represents a multi-dimensional vector composed of cross-table associated features extracted from the archive table, work order table, and settlement table;

[0013] Based on the generalized Mahalanobis distance algorithm, a distance parameter vector is determined according to the associated feature vector, the preset benchmark vector, the preset global covariance matrix, and the user type weight matrix; wherein, the distance parameter vector represents the Mahalanobis distance measure between each archive data in the archive database and the average of historical compliant archives;

[0014] If it is determined that the distance parameter in the distance parameter vector is greater than the first preset threshold, then the file data corresponding to the distance parameter is determined to be abnormal file data.

[0015] If it is determined that the distance parameter in the distance parameter vector is less than or equal to the first preset threshold, then the file data corresponding to the distance parameter is determined to be normal file data.

[0016] In one possible implementation, the archival labeling data is subjected to quality assessment processing based on an interval-type II fuzzy comprehensive evaluation model to obtain archival grade data, including:

[0017] The archival tagging data is encoded to obtain archival coding features; wherein, the archival coding features are used as feature information for quality assessment of completeness, accuracy and standardization.

[0018] Based on the interval type II fuzzy comprehensive evaluation model, the archive coding features are processed to obtain archive level data.

[0019] In one possible implementation, the archive coding features are processed based on an interval-type II fuzzy comprehensive evaluation model to obtain archive grade data, including:

[0020] The missing data is processed by performing a completeness traversal on the archive coding features to obtain a completeness score; the accuracy is tested by performing an accuracy detection on the archive coding features to obtain an accuracy score; and the feature fields in the archive coding features are matched by regular expressions to obtain a normalization score.

[0021] Based on the membership function table of the interval type II fuzzy comprehensive evaluation model, the completeness score, the accuracy score, and the normalization score are processed to obtain a score matrix; wherein, the score matrix represents the fuzzy evaluation matrix obtained after the completeness score, accuracy score, and normalization score are transformed by the membership function.

[0022] The score matrix is ​​weighted based on a preset weight vector to obtain a grade vector; wherein, the grade vector represents the membership vector of each quality data in the archive data to each grade.

[0023] The median of each level in the level vector is used as the centroid representative value for that level; based on the centroid representative value, the archive level data is determined.

[0024] In one possible implementation, an archive anomaly report is generated based on the archive tagging data and the archive level data, including:

[0025] Based on the archival tagging data and the archival grade data, determine the complete archival data;

[0026] The full set of archival data is processed using a multifractal spectrum analysis algorithm to determine the partition function; wherein, the partition function represents the q-th order moment function of the anomaly probability measure of the full set of archival data at different scales;

[0027] Based on the partition function, the singularity index and multifractal spectrum are determined; wherein, the singularity index characterizes the singularity intensity of the erroneous partition distribution's fractal structure; and the multifractal spectrum characterizes the multifractal spectral width and peak characteristics of the anomalous event distribution.

[0028] Anomalies in the archive data are located using the singularity index and the multifractal spectrum, and an archive anomaly report is generated.

[0029] In one possible implementation, obtaining a database of low-voltage customer profiles includes:

[0030] Real-time acquisition of low-voltage customer profile data;

[0031] Based on the normed linear spatial weighted deviation model, the archive data is verified to obtain the verification results.

[0032] If the verification result indicates that the verification failed, an error warning is issued to the user.

[0033] If the verification result indicates that the verification passed, the file data is added to the file database.

[0034] In one possible implementation, the archival data is verified based on a normed linear spatial weighted bias model to obtain verification results, including:

[0035] The archival data is encoded to obtain an archival encoding vector; wherein, the archival encoding vector represents a multi-dimensional numerical vector formed by encoding the values ​​of each field of the archival data;

[0036] Based on the weighted norm algorithm, the file encoding vector and the preset reference vector are processed to obtain the deviation value;

[0037] If the deviation value is determined to be greater than or equal to the second preset threshold, then the verification result of the archive data is determined to be verification failure.

[0038] If the deviation value is determined to be less than the second preset threshold, then the verification result of the archive data is determined to be verified as passed.

[0039] Secondly, embodiments of this application provide a low-voltage customer file anomaly detection device, comprising:

[0040] The acquisition module is used to acquire the low-voltage customer file database; the file database includes at least one file record;

[0041] The processing module is used to perform anomaly detection processing on each piece of archive data in the archive database to obtain anomaly results; and to mark the anomaly results with each piece of archive data to obtain archive mark data;

[0042] The evaluation module is used to perform quality evaluation processing on the archive marking data based on the interval type II fuzzy comprehensive evaluation model to obtain archive grade data;

[0043] The generation module is used to generate an archive anomaly report based on the archive tag data and the archive level data.

[0044] In one possible implementation, the archive data includes an archive table, a work order table, and a settlement table; the processing module is specifically used for:

[0045] Feature extraction processing is performed on the archive table, work order table, and settlement table of each archive data in the archive database to obtain an associated feature vector; wherein, the associated feature vector represents a multi-dimensional vector composed of cross-table associated features extracted from the archive table, work order table, and settlement table;

[0046] Based on the generalized Mahalanobis distance algorithm, a distance parameter vector is determined according to the associated feature vector, the preset benchmark vector, the preset global covariance matrix, and the user type weight matrix; wherein, the distance parameter vector represents the Mahalanobis distance measure between each archive data in the archive database and the average of historical compliant archives;

[0047] If it is determined that the distance parameter in the distance parameter vector is greater than the first preset threshold, then the file data corresponding to the distance parameter is determined to be abnormal file data.

[0048] If it is determined that the distance parameter in the distance parameter vector is less than or equal to the first preset threshold, then the file data corresponding to the distance parameter is determined to be normal file data.

[0049] In one possible implementation, the evaluation module is specifically used for:

[0050] The archival tagging data is encoded to obtain archival coding features; wherein, the archival coding features are used as feature information for quality assessment of completeness, accuracy and standardization.

[0051] Based on the interval type II fuzzy comprehensive evaluation model, the archive coding features are processed to obtain archive level data.

[0052] In one possible implementation, the evaluation module is further configured to:

[0053] The missing data is processed by performing a completeness traversal on the archive coding features to obtain a completeness score; the accuracy is tested by performing an accuracy detection on the archive coding features to obtain an accuracy score; and the feature fields in the archive coding features are matched by regular expressions to obtain a normalization score.

[0054] Based on the membership function table of the interval type II fuzzy comprehensive evaluation model, the completeness score, the accuracy score, and the normalization score are processed to obtain a score matrix; wherein, the score matrix represents the fuzzy evaluation matrix obtained after the completeness score, accuracy score, and normalization score are transformed by the membership function.

[0055] The score matrix is ​​weighted based on a preset weight vector to obtain a grade vector; wherein, the grade vector represents the membership vector of each quality data in the archive data to each grade.

[0056] The median of each level in the level vector is used as the centroid representative value for that level; based on the centroid representative value, the archive level data is determined.

[0057] In one possible implementation, the generation module is specifically used for:

[0058] Based on the archival tagging data and the archival grade data, determine the complete archival data;

[0059] The full set of archival data is processed using a multifractal spectrum analysis algorithm to determine the partition function; wherein, the partition function represents the q-th order moment function of the anomaly probability measure of the full set of archival data at different scales;

[0060] Based on the partition function, the singularity index and multifractal spectrum are determined; wherein, the singularity index characterizes the singularity intensity of the erroneous partition distribution's fractal structure; and the multifractal spectrum characterizes the multifractal spectral width and peak characteristics of the anomalous event distribution.

[0061] Anomalies in the archive data are located using the singularity index and the multifractal spectrum, and an archive anomaly report is generated.

[0062] In one possible implementation, the acquisition module is specifically used for:

[0063] Real-time acquisition of low-voltage customer profile data;

[0064] Based on the normed linear spatial weighted deviation model, the archive data is verified to obtain the verification results.

[0065] If the verification result indicates that the verification failed, an error warning is issued to the user.

[0066] If the verification result indicates that the verification passed, the file data is added to the file database.

[0067] In one possible implementation, the acquisition module is further configured to:

[0068] The archival data is encoded to obtain an archival encoding vector; wherein, the archival encoding vector represents a multi-dimensional numerical vector formed by encoding the values ​​of each field of the archival data;

[0069] Based on the weighted norm algorithm, the file encoding vector and the preset reference vector are processed to obtain the deviation value;

[0070] If the deviation value is determined to be greater than or equal to the second preset threshold, then the verification result of the archive data is determined to be verification failure.

[0071] If the deviation value is determined to be less than the second preset threshold, then the verification result of the archive data is determined to be verified as passed.

[0072] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0073] The memory stores computer-executed instructions;

[0074] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0075] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0076] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0077] This application provides a method, apparatus, electronic device, and medium for detecting anomalies in low-voltage customer archives. It acquires a low-voltage customer archive database containing at least one archive record, performs anomaly detection on each record record in the database to obtain anomaly results, and associates and marks the anomaly results with the corresponding archive records to generate archive marker data representing the associated anomaly results. Then, it introduces an interval-type II fuzzy comprehensive evaluation model to perform quality assessment on the archive marker data, obtaining archive grade data representing the quality level of each archive record. Finally, it combines the archive marker data and archive grade data to generate an archive anomaly report covering anomaly location and risk characteristic analysis in the archive database. This method relies on archive-by-archive detection and marking to achieve precise anchoring of anomaly points. By utilizing the interval-type II fuzzy comprehensive evaluation model to adapt to the uncertainty and fuzziness of low-voltage customer archive data, it effectively improves the objectivity and accuracy of archive quality grading. Furthermore, it generates standardized analysis reports through multi-dimensional data fusion, achieving efficient screening of low-voltage customer archive anomalies, accurate quality grading, and intuitive risk presentation. This provides clear data support and decision-making basis for the rectification, optimization, and standardized management of archive data. Attached Figure Description

[0078] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0079] Figure 1 A flowchart illustrating a method for detecting file anomalies in low-voltage customers provided in this application embodiment. Figure 1 ;

[0080] Figure 2 A flowchart illustrating a method for detecting file anomalies in low-voltage customers provided in this application embodiment. Figure 2 ;

[0081] Figure 3 A schematic diagram of the structure of a low-voltage customer file anomaly detection device provided in this application embodiment. Figure 1 ;

[0082] Figure 4 A schematic diagram of the structure of a low-voltage customer file anomaly detection device provided in this application embodiment. Figure 2 ;

[0083] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0084] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0085] 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 numbers 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.

[0086] Low-voltage customer file management falls under the technical field of electricity marketing information management. It primarily addresses the creation, archiving, verification, and quality control of files for low-voltage electricity customers, including residential, non-residential, and charging facility customers. Related operations typically rely on a marketing file database for centralized management and subsequent review of basic customer information, metering information, settlement information, and related business data.

[0087] Current document review methods typically rely on manual verification combined with basic rule checks. This involves retrieving document data from a database and then having staff compare fields such as name, address, and number item by item based on their experience, or using fixed rules to check for field formats, empty values, and duplicates. This type of method mainly depends on single-field judgments and static rule triggers, and its working principle focuses on screening for surface errors.

[0088] However, with the continuous growth in the number of low-voltage customers and the increasing diversity of user types, the above methods are insufficient to effectively identify anomalies arising across data items and business processes. For example, although a single field may have a correct format, it may have logical inconsistencies, insufficient completeness, or hidden risk characteristics with other related information, leading to inaccurate anomaly localization. Meanwhile, manual review is limited by processing efficiency and subjective judgment, easily causing delays in anomaly identification and making it difficult to conduct unified hierarchical analysis of file quality, thus affecting the efficiency of file management and the targeted nature of subsequent risk handling.

[0089] Therefore, this application proposes a method for detecting file anomalies of low-voltage customers. By acquiring the file database of low-voltage customers, performing anomaly detection processing on each file data and forming file tag data associated with the anomaly results, and then conducting quality assessment on the file tag data based on the interval type II fuzzy comprehensive evaluation model to obtain the corresponding file level data, and then combining the file tag data and the file level data to generate a file anomaly report, so as to realize anomaly location and risk feature analysis.

[0090] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0091] Figure 1 A flowchart illustrating a method for detecting file anomalies in low-voltage customers provided in this application embodiment. Figure 1 ,like Figure 1 As shown, the method includes:

[0092] S101. Obtain the database of low-voltage customer records; the database includes at least one record.

[0093] For example, the archive database is used to store the archive data of low-voltage customers, which is the basic data source for subsequent anomaly detection, quality assessment and report generation.

[0094] The data in the archives may include basic customer information, electricity address information, metering device information, electricity price and type information, settlement account information, connection information, and related fields related to marketing business.

[0095] In practice, the executing entity can be a file processing server deployed in the marketing information management platform, or a business middleware node with data processing capabilities. After the task is started, the executing entity reads low-pressure customer file records from a relational database, distributed data warehouse, or existing marketing file system, and organizes the read results into a data set that can be processed one record at a time.

[0096] In one possible implementation, the archive database can be obtained and organized from one or more business data sources to ensure that related data for the same customer can be merged into the same archive data. In another implementation, the archive database can also be directly extracted from an existing marketing database mirror and batched into archive sets to be tested.

[0097] To ensure a clear data input structure for subsequent processing, this step can perform preprocessing actions after acquiring the archival database. These actions include field standardization, unified null value encoding, date format conversion, administrative division code alignment, and removal of redundant spaces from text fields. For address fields, information such as province, city, district / county, street, and house number can be broken down into structured subfields. For fields with fixed encoding rules, such as document numbers, equipment numbers, and metering point numbers, they can be uniformly converted to string format while retaining leading zeros. For numeric fields, the unit of measurement and currency representation can be converted according to a preset data dictionary.

[0098] The archive database contains at least one archive record, meaning the system is suitable for both single archive verification and batch processing scenarios involving large volumes of customer archives. After reading, the system assigns an internal processing index to each archive record and establishes a mapping relationship between the original archive identifier and the processing index, which is used for subsequent anomaly detection result write-back, quality level association, and report location.

[0099] In one possible embodiment, the database of low-voltage customer records can be accessed through two methods: scheduled task retrieval and manual triggering. Scheduled task retrieval can be executed daily, hourly, or according to business settlement cycles, incrementally extracting new or updated low-voltage customer records from the main marketing database. Manual triggering involves the front-end interaction layer receiving record verification instructions, generating query statements based on filtering conditions such as customer type, power supply unit, transformer area, and record creation time interval, and transmitting the query results to the data processing layer. After receiving the database of low-voltage customer records, the data processing layer caches it in an in-memory data table or a temporary calculation table, providing a stable data access interface for record-by-record anomaly detection.

[0100] By acquiring an archive database containing at least one record, a unified data entry point was established for subsequent processing, ensuring that each low-voltage customer record has an identifiable, traceable, and computable data carrier.

[0101] S102. Perform anomaly detection processing on each piece of archival data in the archival database to obtain anomaly results; and mark the anomaly results with each piece of archival data to obtain archival marked data.

[0102] Among them, the archive tag data represents the archive data that has been associated with abnormal results.

[0103] For example, the anomaly detection process in this step is used to identify anomalies in the archival data and obtain anomaly results, and is the core processing step of the method; the archival tag data is used to characterize the archival data associated with the anomaly results, which facilitates subsequent quality assessment and anomaly location.

[0104] In practice, the system iterates through each record in the archive database using a pre-established processing index, constructing a detection input for each record. This detection input includes at least the basic field values, the relationships between fields, and the business rule matching results.

[0105] Anomaly detection and handling can proceed by first performing basic consistency checks, then performing correlation logic checks, and finally performing deviation analysis. Basic consistency checks include determining whether required fields are missing, whether the length of encoded fields meets constraints, whether date fields satisfy chronological order, and whether the content of related fields is consistent.

[0106] The correlation logic verification includes determining whether the customer name matches the customer type, whether the electricity address matches the power supply unit's jurisdiction, whether the metering device status matches the settlement status, and whether the meter installation and connection time matches the filing time. Deviation analysis compares the current file with a group of files in the same region, customer category, electricity price type, or connection method to identify file records that, although the field format is correct, have abnormal combination relationships.

[0107] In one possible embodiment, the system extracts feature items for each piece of archival data and forms anomaly detection features. These feature items may include field integrity features, field normalization features, field consistency features, address structure integrity features, numbering rule compliance features, and customer profile deviation features. For text fields, standardized expressions can be generated first through word segmentation, dictionary mapping, or standard address database matching; for categorical fields, they can be converted to discrete codes; for numerical fields, normalization can be used to unify the scale. Based on this, rule-based detection and / or statistical detection methods can be used to output anomaly results. The rule-based detection part can output anomaly type codes, such as missing anomalies, conflict anomalies, duplication anomalies, logical anomalies, and out-of-bounds anomalies; the statistical detection part can identify the degree of deviation of the current archive relative to the reference archive set based on sample distribution and write the corresponding deviation dimension into the anomaly results.

[0108] After outputting abnormal results, the system marks the abnormal results with each corresponding file data to form file mark data.

[0109] Specifically, the system appends a set of anomaly marker fields to the original archive records. This set of fields includes at least an anomaly status bit, an anomaly type list, an anomaly severity, an anomaly field location, a detection timestamp, and a detection batch number. The anomaly status bit indicates whether the current archive contains an anomaly; the anomaly type list records multiple anomaly categories that the archive contains; the anomaly severity can be determined based on the number of anomalies and the impact range of key fields; and the anomaly field location is used to precisely locate the problematic field name or field group. After the marking process, each archive record forms a composite record structure of original archive content and anomaly association information, thus obtaining the archive marking data. For archive data where no anomalies are detected, the system also writes a normal identifier and an empty anomaly list to maintain consistency in the input structure for subsequent quality assessments.

[0110] This step establishes a one-to-one correspondence between abnormal results and archived data, enabling subsequent processing to directly access specific abnormal content without having to revisit the original detection process. For cross-field anomalies, the system can record multiple participating fields at the anomaly field location.

[0111] By performing anomaly detection processing on each archive data and generating archive tag data, not only can surface errors be identified, but also hidden inconsistencies between fields and business processes can be expressed in a structured way, thus providing quantifiable input for subsequent quality level calculations.

[0112] S103. Based on the interval type II fuzzy comprehensive evaluation model, the quality assessment of the archival marking data is performed to obtain the archival grade data.

[0113] Among them, the archival grade data represents the quality level of each archival data in the archival database.

[0114] For example, the interval type II fuzzy comprehensive evaluation model is used to perform quality assessment on archival tag data to obtain archival grade data; the archival grade data is used to characterize the quality grade of each archival data in the archival database and is the output of the quality assessment.

[0115] In practice, the system extracts one or more evaluation factors for quality assessment from the archival tagging data and constructs a set of evaluation indicators. The system converts these evaluation factors into membership degree inputs according to a unified dimension and performs a comprehensive evaluation on a per-archive basis.

[0116] The interval-type fuzzy comprehensive evaluation model processes data by establishing multiple membership functions for each evaluation factor. Each level corresponds to an interval-based membership degree, rather than a single precise value. Specifically, the system can set five evaluation levels: high quality, high quality, average quality, low quality, and low quality. For each evaluation factor, upper and lower membership functions are defined to represent the membership intervals of that factor to each level under different sample fluctuations. After completing the interval membership mapping for each evaluation factor, the system performs fuzzy synthesis based on pre-set weights to obtain the comprehensive evaluation result for each archive. The final archive level data can be output through defuzzification or the maximum membership principle.

[0117] In one possible embodiment, the system can output an archive grade label and quality score based on the comprehensive evaluation results. If an archive has multiple anomalies, its corresponding evaluation result can shift to a lower grade, ultimately generating lower quality grade data; if there are only minor anomalies and the other major evaluation results are normal, its comprehensive evaluation result can fall into a higher grade range.

[0118] By employing an interval-type fuzzy comprehensive evaluation model, this step can address the uncertainties arising from the degree of anomalies, data fluctuations, and unclear indicator boundaries in archival quality evaluation, enabling different archives to achieve a unified grade expression under conditions of multiple indicators and multiple anomalies. The obtained archival grade data can include archival identification, quality grade label, quality score, and evaluation time.

[0119] S104. Generate an archive anomaly report based on archive tag data and archive level data.

[0120] Among them, the archive anomaly report is a report that identifies anomalies and analyzes risk characteristics in the archive database.

[0121] For example, the archive anomaly report is used to characterize the results of anomaly location and risk characteristic analysis of the archive database, and is the final output report.

[0122] In practice, the system associates archive tagging data with archive level data according to archive identifiers and constructs a report dataset. This report dataset includes at least customer identification information, anomaly status, anomaly type, anomaly field location, anomaly severity, quality level, quality score, and statistical dimensions such as the power supply unit, customer category, and filing time. Subsequently, the system generates anomaly reports for management and handling according to a preset report template. The report template can include both detail pages and summary pages. The detail pages display the location and content of each anomaly archive, while the summary pages statistically analyze the distribution across different regions, customer categories, anomaly types, and quality levels.

[0123] In the specific generation process, the system can first perform anomaly location processing, that is, map each anomaly file to a specific field, specific business process and specific customer object according to the location of the anomaly field; then perform risk feature analysis processing, that is, perform statistical summary around the combination of anomaly types, the stratification of anomaly severity and the distribution of quality levels.

[0124] In one possible embodiment, the system can cross-aggregate archive tagging data and archive level data according to dimensions such as power supply unit, transformer area, customer type, and archive batch to obtain a multi-dimensional risk profile. For frequently occurring abnormal patterns, the system can mark them as key targets for governance; for areas or business processes with an abnormally high proportion of low-quality archives, the system can add warning indicators to the reports.

[0125] The generated archive anomaly reports can be output as structured electronic reports, database result tables, or visual pages. In structured electronic reports, each row corresponds to one archive record or one type of anomaly statistical item; in database result tables, a main report table and detailed report tables can be stored. The main table records the report batch, statistical range, and summary indicators, while the detailed tables record specific abnormal archives and their level results; in visual pages, lists, bar charts, heatmaps, or distribution maps can be used to display the anomaly location and risk characteristics, but their underlying data still originates from the same archive tagging data and archive level data. After the report is generated, the task execution layer can write the report to the report library, push it to the front-end interaction layer, or output it in a file format for business personnel to download, including the generation time, data range, and task number, ensuring that the report results are auditable and traceable.

[0126] By simultaneously utilizing the location information of abnormal results and the hierarchical information of quality levels, the report no longer stops at listing errors in a single field, but can reflect the specific location of abnormalities in the archive database, the combination relationship of abnormalities, and the distribution of risk levels.

[0127] This application provides a method for detecting anomalies in low-voltage customer archives. It involves acquiring a database of low-voltage customer archives containing at least one archive record, performing anomaly detection on each record to obtain anomaly results, and associating these results with corresponding archive records to generate archive tag data representing the associated anomaly results. Then, an interval-type II fuzzy comprehensive evaluation model is introduced to perform quality assessment on the archive tag data, obtaining archive grade data representing the quality level of each archive record. Finally, combining the archive tag data and archive grade data, an archive anomaly report is generated, covering anomaly location and risk characteristic analysis in the archive database. This method achieves precise anchoring of anomalies by detecting and tagging each archive record, and utilizes the interval-type II fuzzy comprehensive evaluation model to adapt to the uncertainty and fuzziness of low-voltage customer archive data, effectively improving the objectivity and accuracy of archive quality grading. Furthermore, by fusing multi-dimensional data to generate standardized analysis reports, it achieves efficient screening of low-voltage customer archive anomalies, accurate quality grading, and intuitive risk presentation, providing clear data support and decision-making basis for the rectification, optimization, and standardized management of archive data.

[0128] Figure 2 A flowchart illustrating a method for detecting file anomalies in low-voltage customers provided in this application embodiment. Figure 2 ,like Figure 2 As shown, in this embodiment... Figure 2 Based on the examples, a method for detecting anomalies in the files of low-voltage customers is described in detail. The method includes:

[0129] S201. Obtain the database of low-voltage customer records; the database includes at least one record.

[0130] For example, this step can be referred to step S101, and will not be repeated here.

[0131] In one example, the system acquires low-voltage customer profile data in real time; it performs verification processing on the profile data based on a normed linear spatial weighted deviation model to obtain the verification result; if the verification result indicates that the verification failed, an abnormal warning is issued to the user; if the verification result indicates that the verification passed, the profile data is added to the profile database.

[0132] The archive data refers to low-voltage customer archive objects collected in real time before being entered into the database and awaiting verification. This includes customer identification, electricity usage attributes, metering-related information, and settlement-related information. The normed linear space weighted deviation model is a mathematical model for verifying the consistency and integrity of the archive data. The system encodes the archive data into an archive code vector and compares it with a preset benchmark vector using a weighted norm to obtain the deviation value. The verification result indicates whether the archive data meets the database entry conditions. If the deviation value exceeds a threshold, the verification fails; if the deviation value does not exceed the threshold, the verification passes. Anomaly warnings are used to alert the user when verification fails, indicating that the current archive has missing fields, logical conflicts, or inconsistent associations.

[0133] In its implementation, the system connects to a front-end document collection interface to capture new, modified, or supplementary data from low-voltage customers in real time. After the data enters the cache, it performs field standardization processing, encoding the name, address, number, contact person, and business-related fields into a vector representation with a unified dimension. Subsequently, the system calls a normed linear space weighted deviation model to calculate the weighted deviation between this vector and a preset benchmark vector obtained from historical compliance documents. The weights correspond to field importance, business type, and risk sensitivity. The deviation value output by the model is then compared with a second preset threshold to form a verification result. When the verification result is a failure, the system immediately sends an exception warning to the user terminal and retains the corresponding exception reason identifier. When the verification result is a success, the system writes the document data to the document database and synchronously updates the index and association mapping relationship. The model used can directly measure the deviation of multi-field combinations. In practical applications, this model can also choose other equivalent normed space representation methods, which are not limited in this application.

[0134] This allows the data entry process to be synchronized with quality assessment, enabling abnormal data to be identified and alerted before being written, while compliant data can be directly archived. This improves the accuracy and efficiency of data entry verification and enhances the consistency and availability of data in the archive database.

[0135] In one example, the archive data is encoded to obtain an archive encoding vector; based on the weighted norm algorithm, the archive encoding vector is processed with a preset benchmark vector to obtain a deviation value; if the deviation value is determined to be greater than or equal to a second preset threshold, the verification result of the archive data is determined to be verification failure; if the deviation value is determined to be less than the second preset threshold, the verification result of the archive data is determined to be verification success.

[0136] Among them, the archive coding vector represents the multi-dimensional numerical vector formed by encoding the values ​​of each field of the archive data.

[0137] For example, after receiving archival data, the system first encodes text fields, discretely numbers enumerated fields, and normalizes numerical fields, then combines the encoding results into an archival encoding vector according to the field order. A preset baseline vector can be generated from the statistical mean of historical compliant archives, business rule baseline values, or standard template vectors to adapt to different archive types. Subsequently, the system performs a weighted summation or weighted norm operation on the dimension-by-dimensional difference between the archival encoding vector and the preset baseline vector. The weights corresponding to different fields are preset according to the importance of the fields in the archival verification, so that deviations in key fields have a more significant impact on the deviation value. When the deviation value reaches or exceeds a second preset threshold, the system outputs a verification failure result and marks the archival data as data to be reviewed; when the deviation value is less than the second preset threshold, the system outputs a verification pass result and allows the archival data to enter subsequent archiving processing.

[0138] This approach vectorizes archival data and incorporates a weighted norm metric, enabling the verification process to uniformly handle differences across multiple fields and incorporating field importance into the judgment criteria. This results in verification outcomes that better align with archival compliance characteristics. Outputting pass / fail results based on threshold comparisons ensures clear and consistent archival verification judgments and provides direct input for subsequent anomaly identification and archival quality control.

[0139] S202. Perform anomaly detection processing on each piece of archival data in the archival database to obtain anomaly results; and mark the anomaly results with each piece of archival data to obtain archival marked data.

[0140] For example, this step can be referred to step S102, and will not be described again.

[0141] In one example, feature extraction is performed on the file table, work order table, and settlement table of each file in the file database to obtain an associated feature vector. Based on the generalized Mahalanobis distance algorithm, a distance parameter vector is determined according to the associated feature vector, a preset benchmark vector, a preset global covariance matrix, and a user type weight matrix. If the distance parameter in the distance parameter vector is greater than a first preset threshold, the file data corresponding to that distance parameter is determined to be abnormal file data; if the distance parameter in the distance parameter vector is less than or equal to the first preset threshold, the file data corresponding to that distance parameter is determined to be normal file data.

[0142] The archival data includes archival tables, work order tables, and settlement tables; the association feature vector represents a multi-dimensional vector composed of cross-table association features extracted from the archival tables, work order tables, and settlement tables; and the distance parameter vector represents the Mahalanobis distance metric between each archival data point in the archival database and the historical average of compliant archives.

[0143] In its implementation, the system first aligns the fields of the file table, work order table, and settlement table corresponding to the same file data, and encodes information that reflects cross-table consistency, logical correlation, and completeness to form a correlation feature vector. Then, it calculates the center deviation of this vector against a preset benchmark vector, and corrects the correlation of different dimensions using a preset global covariance matrix. Simultaneously, it weights the feature contributions of different user types according to a user type weight matrix to obtain a distance parameter vector. Finally, it compares each distance parameter in the distance parameter vector with a first preset threshold to output abnormal or normal file data. If the feature encoding method used is discrete field concatenation, numerical field normalization, and category field vectorization, the correlation differences between different source tables can be uniformly mapped to the same feature space. If the generalized Mahalanobis distance calculation module is implemented using matrix operations, the distance parameter corresponding to each file data can be directly output, facilitating subsequent labeling and summarization. In practical applications, the encoding module, distance calculation module, and threshold comparison module can also adopt other models or implementation methods, which are not limited in this application.

[0144] This allows cross-table logical anomalies to be expressed uniformly, and uses the generalized Mahalanobis distance algorithm to measure deviations in multi-dimensional features. Combined with the user type weight matrix to correct differences in different business types, it enables anomaly judgment to be based on the statistical basis of the average of historical compliance files, thereby outputting more consistent anomaly results.

[0145] By adopting this specific implementation method, hidden problems such as cross-table inconsistencies, business link breaks, and settlement association anomalies can be identified when the surface fields of the archives are correct. It also provides a unified quantitative standard for distinguishing between abnormal and normal archives, thereby providing a stable data basis for subsequent archive marking, quality assessment, and archive anomaly report generation.

[0146] S203. Encode the archival marking data to obtain archival coding features; based on the interval type II fuzzy comprehensive evaluation model, process the archival coding features to obtain archival grade data.

[0147] Among them, the archival coding features are used to assess the completeness, accuracy, and standardization of the quality assessment.

[0148] In its implementation, after receiving the archive tagging data, the system first discretizes and encodes the field information, anomaly association information, and tagging status information, converting text, enumeration values, and Boolean states into numerical features. Then, it generates completeness, accuracy, and standardization codes according to preset mapping rules. The coding rules record missing fields as low values, content matching as intermediate values, and format compliance as high values, incorporating the weights of different archive types into the feature vector to form archive coding features. This feature vector is then input into a type-II fuzzy comprehensive evaluation model. The model performs fuzzy inference on each dimension based on membership intervals, evaluation factor weights, and grade thresholds, and then performs comprehensive aggregation, ultimately outputting archive grade data corresponding to archive quality. The adopted coding method allows anomaly association information to be uniformly included in the evaluation input. In practical applications, this coding rule can also be extended to accommodate different business criteria; this application does not limit this extension.

[0149] This processing method involves first structuring and encoding the archival data, then using a type-II fuzzy comprehensive evaluation model to complete the comprehensive assessment. This maps completeness, accuracy, and standardization into a unified grade result, providing a stable quality basis for the subsequent generation of archival anomaly reports. Because the encoded features are computable and comparable, the archival quality assessment results can maintain a consistent grading standard, thereby improving the output stability and consistency of archival grade data.

[0150] In one example, missing data is processed to obtain a completeness score for the archive coding features; accuracy is tested to obtain an accuracy score for the archive coding features; regular expression matching is performed on the feature fields in the archive coding features to obtain a normalization score; based on the membership function table of the interval type II fuzzy comprehensive evaluation model, the completeness score, accuracy score, and normalization score are processed to obtain a score matrix; the score matrix is ​​weighted based on a preset weight vector to obtain a grade vector; the median of each grade in the grade vector is used as the centroid representative value of that grade; and the archive grade data is determined based on the centroid representative value.

[0151] Among them, the score matrix represents the fuzzy evaluation matrix obtained after the completeness score, accuracy score and normality score are transformed by the membership function; the rank vector represents the membership vector of each archival data in the archival data to each rank.

[0152] In its implementation, the system first performs a missing value traversal on the archive coding features, checking each field for null, unfilled, or unparseable values, and calculates the ratio of complete fields to the total number of fields as a completeness score. Next, it performs an accuracy check on the archive coding features, comparing field values ​​with baseline information, related table data, or business rules in the archive database, and generating an accuracy score based on the matching results. Then, it performs regular expression matching on the feature fields in the archive coding features, comparing the number, date, address, and code fields with preset format templates item by item, and generating a standardization score based on the degree of matching success. Using the membership function table of the interval type-two fuzzy comprehensive evaluation model, the system converts the above three types of scores into membership values ​​for corresponding levels and organizes them into a score matrix. Then, it performs a weighted operation on the score matrix according to a preset weight vector to obtain a level vector reflecting the degree of membership at each level. Further, the system calculates the median of the membership distribution for each level in the level vector and uses the median as the centroid representative value for that level, ultimately determining the archive level data based on the centroid representative value.

[0153] This processing method evaluates missing data, accuracy, and standardization in parallel, and combines fuzzy membership, weight allocation, and center of gravity determination to complete the grade output. This enables the quality of archives to be graded and characterized on a unified scale, thereby forming archive grade data that can be directly used for subsequent governance and correlation analysis.

[0154] S204. Generate an archive anomaly report based on archive tag data and archive level data.

[0155] For example, this step can be referred to step S104, and will not be described again.

[0156] In one example, the entire set of archival data is determined based on archival tagging data and archival grade data; a multifractal spectrum analysis algorithm is used to process the entire set of archival data to determine the partition function; based on the partition function, the singularity index and multifractal spectrum are determined; and anomalies in the archival data are located using the singularity index and multifractal spectrum, and an archival anomaly report is generated.

[0157] Among them, the partition function represents the q-th order moment function of the anomaly probability measure of the full archive data at different scales; the singularity index represents the singularity intensity of the partial geometric structure of the error distribution; and the multifractal spectrum represents the multifractal spectral width and peak characteristics of the distribution of anomalous events.

[0158] In practical processing, the system first correlates and summarizes the anomaly markers in the archival tagging data with the grade results in the archival grade data to form a complete set of archival data covering all archives to be analyzed. This data is then organized into a computable data sequence or matrix, which can be arranged by archive number, anomaly type, grade value, and associated fields for subsequent scale division and statistical analysis. Subsequently, the system extracts anomaly measures from the complete archival data across multiple scale windows and performs moment calculations on the measure distribution under different q-values ​​to obtain the partition function. Based on this, the system obtains the singularity index by fitting the scale response relationship of the partition function and further constructs a multifractal spectrum to characterize the degree of aggregation and distribution differences of anomalous events in different local regions. The system then jointly maps the local mutation locations corresponding to the singularity index with the spectral width and peak information corresponding to the multifractal spectrum to determine the archival segment, field association relationship, or quality level distribution of the anomalous data, and outputs an archival anomaly report containing the anomaly location, anomaly characteristics, and distribution intensity.

[0159] In this implementation, by leveraging the unified construction of the entire archive data and multifractal spectrum analysis, discrete anomaly markers and ranking results can be transformed into quantifiable distribution characteristics. Singularity indices are used to pinpoint local anomaly clusters, while multifractal spectrum analysis characterizes the complexity of anomaly distributions. The combination of these two methods generates reports with both location and distribution description capabilities. Therefore, the archive anomaly report not only reflects the location of anomaly data but also the scale variation characteristics of anomalies across the entire archive, thus providing support for subsequent archive verification, risk classification and handling, and quality management.

[0160] This application provides a method for detecting anomalies in low-voltage customer archives. It involves acquiring a database of low-voltage customer archives containing at least one record, performing anomaly detection on each record and marking it accordingly to generate archive marking data. This marked data is then encoded to extract three types of archive coding features for quality assessment: completeness, accuracy, and standardization. An interval-type II fuzzy comprehensive evaluation model is then introduced to quantitatively analyze and comprehensively evaluate these features, resulting in archive grade data. Finally, the archive marking data and archive grade data are combined to generate a standardized archive anomaly report. This systematic technical approach utilizes feature coding to achieve multi-dimensional quantitative representation of archive quality and leverages the interval-type II fuzzy comprehensive evaluation model to adapt to the fuzziness and uncertainty of low-voltage customer archive data, enhancing the objectivity and fault tolerance of anomaly detection. This method achieves automated, multi-dimensional, accurate detection and hierarchical management of low-voltage customer archive anomalies, effectively reducing the workload and omission rate of manual verification. It clearly presents the locations of archive anomalies and the overall quality level, providing intuitive and reliable data support and decision-making basis for the standardized governance and continuous improvement of data quality in low-voltage customer archives.

[0161] Figure 3 A schematic diagram of the structure of a low-voltage customer file anomaly detection device provided in this application embodiment. Figure 1 ,like Figure 3 As shown, the low-voltage customer file anomaly detection device 30 provided in this embodiment includes:

[0162] Module 301 is used to acquire the low-voltage customer record database; the record database includes at least one record.

[0163] Processing module 302 is used to perform anomaly detection processing on each piece of archival data in the archival database to obtain anomaly results; and to mark the anomaly results with each piece of archival data to obtain archival marked data;

[0164] Evaluation module 303 is used to perform quality evaluation processing on archival marker data based on the interval type II fuzzy comprehensive evaluation model to obtain archival grade data;

[0165] The generation module 304 is used to generate an archive anomaly report based on archive tag data and archive level data.

[0166] This embodiment provides a low-voltage customer file anomaly detection device, which can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0167] Figure 4 A schematic diagram of the structure of a low-voltage customer file anomaly detection device provided in this application embodiment. Figure 2 ,like Figure 4 As shown, the low-voltage customer file anomaly detection device 40 provided in this embodiment includes:

[0168] Module 401 is used to acquire the low-voltage customer record database; the record database includes at least one record.

[0169] Processing module 402 is used to perform anomaly detection processing on each piece of archival data in the archival database to obtain anomaly results; and to mark the anomaly results with each piece of archival data to obtain archival marked data;

[0170] Evaluation module 403 is used to perform quality evaluation processing on archival marker data based on the interval type II fuzzy comprehensive evaluation model to obtain archival grade data;

[0171] Module 404 is used to generate an archive anomaly report based on archive tag data and archive level data.

[0172] In one possible implementation, the archive data includes an archive table, a work order table, and a settlement table; the processing module 402 is specifically used for:

[0173] Feature extraction is performed on the archive table, work order table, and settlement table of each archive data in the archive database to obtain the associated feature vector; wherein, the associated feature vector represents a multi-dimensional vector composed of cross-table associated features extracted from the archive table, work order table, and settlement table;

[0174] Based on the generalized Mahalanobis distance algorithm, the distance parameter vector is determined according to the associated feature vector, the preset benchmark vector, the preset global covariance matrix, and the user type weight matrix; wherein, the distance parameter vector represents the Mahalanobis distance measure between each archive data in the archive database and the average of historical compliant archives.

[0175] If it is determined that the distance parameter in the distance parameter vector is greater than the first preset threshold, then the file data corresponding to the distance parameter is determined to be abnormal file data.

[0176] If the distance parameter in the distance parameter vector is determined to be less than or equal to the first preset threshold, then the file data corresponding to that distance parameter is determined to be normal file data.

[0177] In one possible implementation, the evaluation module 403 is specifically used for:

[0178] The archival tagging data is encoded to obtain archival coding features; among them, the archival coding features are used to provide characteristic information for the completeness, accuracy and standardization of quality assessment.

[0179] Based on the interval type II fuzzy comprehensive evaluation model, the archival coding features are processed to obtain archival grade data.

[0180] In one possible implementation, the evaluation module 403 is further configured to:

[0181] Completeness score is obtained by performing missing traversal processing on the archive coding features; accuracy score is obtained by performing accuracy detection processing on the archive coding features; and normalization score is obtained by performing regular expression matching processing on the feature fields in the archive coding features.

[0182] Based on the membership function table of the interval type II fuzzy comprehensive evaluation model, the completeness score, accuracy score, and normality score are processed to obtain the score matrix; where the score matrix represents the fuzzy evaluation matrix obtained after the completeness score, accuracy score, and normality score are transformed by the membership function.

[0183] The score matrix is ​​weighted based on a preset weight vector to obtain a grade vector; where the grade vector represents the membership vector of each quality data in the archive data to each grade.

[0184] The median of each level in the level vector is used as the centroid representative value for that level; based on the centroid representative value, the archive level data is determined.

[0185] In one possible implementation, the generation module 404 is specifically used for:

[0186] Based on archival marking data and archival grade data, determine the total amount of archival data;

[0187] A multifractal spectrum analysis algorithm is used to process the full set of archival data and determine the partition function; the partition function represents the q-th moment function of the anomaly probability measure of the full set of archival data at different scales.

[0188] Based on the partition function, the singularity index and multifractal spectrum are determined; whereby the singularity index characterizes the singularity intensity of the erroneous distribution of the multifractal structure; and the multifractal spectrum characterizes the multifractal spectral width and peak characteristics of the distribution of anomalous events.

[0189] Anomalies in archival data are located using singularity indices and multifractal spectra, and an anomaly report is generated.

[0190] In one possible implementation, the acquisition module 401 is specifically used for:

[0191] Real-time acquisition of low-voltage customer profile data;

[0192] Based on the normed linear spatial weighted deviation model, the archival data is verified to obtain the verification results.

[0193] If the verification result indicates that the verification failed, an exception warning will be issued to the user.

[0194] If the verification result indicates that the verification passed, the file data will be added to the file database.

[0195] In one possible implementation, the acquisition module 401 is further configured to:

[0196] The archival data is encoded to obtain an archival encoding vector; the archival encoding vector represents a multi-dimensional numerical vector formed by encoding the values ​​of each field of the archival data.

[0197] Based on the weighted norm algorithm, the file encoding vector and the preset benchmark vector are processed to obtain the deviation value;

[0198] If the deviation value is determined to be greater than or equal to the second preset threshold, then the verification result of the archive data is determined to be verification failure.

[0199] If the deviation value is determined to be less than the second preset threshold, then the verification result of the archive data is determined to be verified as passed.

[0200] This embodiment provides a low-voltage customer file anomaly detection device, which can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0201] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 5 As shown, the electronic device 50 provided in this embodiment includes at least one processor 501 and a memory 502. Optionally, the electronic device 50 further includes a communication component 503. The processor 501, memory 502, and communication component 503 are connected via a bus 504.

[0202] In a specific implementation, at least one processor 501 executes computer execution instructions stored in memory 502, causing at least one processor 501 to perform the above-described method.

[0203] The specific implementation process of processor 501 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0204] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0205] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0206] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0207] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0208] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0209] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0210] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0211] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0212] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0213] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0214] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0215] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0216] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for detecting anomalies in the files of low-voltage customers, characterized in that, include: Obtain a database of low-voltage customer records; the database includes at least one record. Anomaly detection processing is performed on each piece of archival data in the archival database to obtain anomaly results; and the anomaly results are then marked with each piece of archival data to obtain archival marked data; wherein, the archival marked data represents the archival data associated with the anomaly results; Based on the interval type II fuzzy comprehensive evaluation model, the archival tag data is subjected to quality assessment processing to obtain archival grade data; wherein, the archival grade data represents the quality grade of each archival data in the archival database; Based on the archive tagging data and the archive level data, an archive anomaly report is generated; wherein, the archive anomaly report represents the anomaly location and risk characteristic analysis report of the archive database.

2. The method according to claim 1, characterized in that, The archive data includes archive forms, work order forms, and settlement forms; Anomaly detection processing is performed on each piece of data in the archive database to obtain anomaly results, including: Feature extraction processing is performed on the archive table, work order table, and settlement table of each archive data in the archive database to obtain an associated feature vector; wherein, the associated feature vector represents a multi-dimensional vector composed of cross-table associated features extracted from the archive table, work order table, and settlement table; Based on the generalized Mahalanobis distance algorithm, a distance parameter vector is determined according to the associated feature vector, the preset benchmark vector, the preset global covariance matrix, and the user type weight matrix; wherein, the distance parameter vector represents the Mahalanobis distance measure between each archive data in the archive database and the average of historical compliant archives; If it is determined that the distance parameter in the distance parameter vector is greater than the first preset threshold, then the file data corresponding to the distance parameter is determined to be abnormal file data. If it is determined that the distance parameter in the distance parameter vector is less than or equal to the first preset threshold, then the file data corresponding to the distance parameter is determined to be normal file data.

3. The method according to claim 1, characterized in that, Based on the interval-type II fuzzy comprehensive evaluation model, the archival marking data is subjected to quality assessment processing to obtain archival grade data, including: The archival tagging data is encoded to obtain archival coding features; wherein, the archival coding features are used as feature information for quality assessment of completeness, accuracy and standardization. Based on the interval type II fuzzy comprehensive evaluation model, the archive coding features are processed to obtain archive level data.

4. The method according to claim 3, characterized in that, Based on the interval-type fuzzy comprehensive evaluation model, the archive coding features are processed to obtain archive grade data, including: The missing data is processed by performing a completeness traversal on the archive coding features to obtain a completeness score; the accuracy is tested by performing an accuracy detection on the archive coding features to obtain an accuracy score; and the feature fields in the archive coding features are matched by regular expressions to obtain a normalization score. Based on the membership function table of the interval type II fuzzy comprehensive evaluation model, the completeness score, the accuracy score, and the normalization score are processed to obtain a score matrix; wherein, the score matrix represents the fuzzy evaluation matrix obtained after the completeness score, accuracy score, and normalization score are transformed by the membership function. The score matrix is ​​weighted based on a preset weight vector to obtain a grade vector; wherein, the grade vector represents the membership vector of each quality data in the archive data to each grade. The median of each level in the level vector is used as the centroid representative value for that level; based on the centroid representative value, the archive level data is determined.

5. The method according to claim 1, characterized in that, Based on the archive tagging data and the archive level data, an archive anomaly report is generated, including: Based on the archival tagging data and the archival grade data, determine the complete archival data; The full set of archival data is processed using a multifractal spectrum analysis algorithm to determine the partition function; wherein, the partition function represents the q-th order moment function of the anomaly probability measure of the full set of archival data at different scales; Based on the partition function, the singularity index and multifractal spectrum are determined; wherein, the singularity index characterizes the singularity intensity of the erroneous partition distribution's fractal structure; and the multifractal spectrum characterizes the multifractal spectral width and peak characteristics of the anomalous event distribution. Anomalies in the archive data are located using the singularity index and the multifractal spectrum, and an archive anomaly report is generated.

6. The method according to any one of claims 1-5, characterized in that, Obtain a database of low-voltage customer profiles, including: Real-time acquisition of low-voltage customer profile data; Based on the normed linear spatial weighted deviation model, the archive data is verified to obtain the verification results. If the verification result indicates that the verification failed, an error warning is issued to the user. If the verification result indicates that the verification passed, the file data is added to the file database.

7. The method according to claim 6, characterized in that, Based on the normed linear spatial weighted deviation model, the archival data is verified to obtain the verification results, including: The archival data is encoded to obtain an archival encoding vector; wherein, the archival encoding vector represents a multi-dimensional numerical vector formed by encoding the values ​​of each field of the archival data; Based on the weighted norm algorithm, the file encoding vector and the preset reference vector are processed to obtain the deviation value; If the deviation value is determined to be greater than or equal to the second preset threshold, then the verification result of the archive data is determined to be verification failure. If the deviation value is determined to be less than the second preset threshold, then the verification result of the archive data is determined to be verified as passed.

8. A device for detecting file anomalies in low-voltage customers, characterized in that, include: The acquisition module is used to acquire the low-voltage customer file database; the file database includes at least one file record; The processing module is used to perform anomaly detection processing on each piece of archive data in the archive database to obtain anomaly results; and to mark the anomaly results with each piece of archive data to obtain archive mark data; The evaluation module is used to perform quality evaluation processing on the archive marking data based on the interval type II fuzzy comprehensive evaluation model to obtain archive grade data; The generation module is used to generate an archive anomaly report based on the archive tag data and the archive level data.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

Images