Method, device, medium and product for detecting abnormality of multi-account power consumption data

Abstract

This invention discloses a method, apparatus, medium, and product for anomaly detection of multi-account electricity consumption data. The method involves: collecting electricity consumption data from each authorized account in an electricity account set; aligning the data to obtain several unified-caliber electricity consumption records and generating corresponding tag sets and indicator sets; performing initial anomaly analysis on each indicator set to obtain initial anomaly information; correcting the initial anomaly information using the tag sets to obtain a correction result; scoring each anomaly indicator item in the correction result using preset rules to obtain a rule score; inputting each anomaly indicator item into a preset anomaly detection model to obtain a model score; weighted and fused the rule scores and model scores corresponding to each anomaly indicator item to obtain a fusion score; and generating the anomaly detection result for the current user's multi-account electricity consumption data based on the fusion score. This invention solves the problem of unified anomaly detection for multi-account electricity consumption data.

Description

Technical Field

[0001] This invention relates to the field of power data processing technology, and in particular to a method, apparatus, medium, and product for detecting anomalies in multi-account electricity consumption data. Background Technology

[0002] With the deepening of the digital transformation of the power system, the granularity and real-time nature of electricity data collection are constantly improving, and power companies and users have an increasing demand for refined management and anomaly monitoring of electricity consumption behavior. For users in scenarios such as multiple households per person, multiple meters per household, joint management by multiple family members, and multiple power stations / shops for enterprises, it is often necessary to frequently switch between multiple electricity accounts to view information such as electricity consumption, electricity consumption structure, and tiered progress for each account. Unified anomaly detection of electricity consumption data from multiple accounts can promptly identify problems such as sudden changes in electricity consumption, imbalances in electricity consumption structure, and metering deviations. This helps users identify equipment failures, optimize electricity consumption strategies, and prevent electricity consumption risks, while also assisting power companies in improving service quality and grid operation stability.

[0003] In existing technologies, anomaly detection of electricity consumption data typically employs time-series curve comparison methods. However, these methods have significant drawbacks: firstly, they only analyze single electricity accounts and cannot handle cross-account comparative analysis and unified anomaly detection in multi-account scenarios; secondly, electricity consumption data from different accounts differ in field naming, units of measurement, and sub-item structures. Existing technologies lack a unified mechanism for processing these differences in data sources and structures, making direct comparison and analysis of multi-account data impossible, and resulting in anomaly detection results that fail to accurately reflect the true electricity consumption situation in multi-account scenarios. Therefore, how to achieve unified anomaly detection for multi-account electricity consumption data with differences in data sources and structures has become an urgent technical problem to be solved. Summary of the Invention

[0004] This invention provides a method, apparatus, medium, and product for detecting anomalies in multi-account electricity consumption data, which can solve the problem of difficulty in uniformly detecting anomalies in multi-account electricity consumption data when the sources and structures of multi-account electricity consumption data are different.

[0005] In a first aspect, an embodiment of the present invention provides a method for detecting anomalies in multi-account electricity consumption data, comprising: Collect electricity consumption data from each authorized account in the electricity account set, wherein the electricity account set includes all the authorized accounts of the current user; The electricity consumption data are aligned to obtain several unified electricity consumption records, and a corresponding tag set and indicator set are generated based on each unified electricity consumption record. An initial anomaly analysis is performed on each set of indicators to obtain initial anomaly information. The initial anomaly information is then corrected using each set of tags to obtain a correction result. Each anomaly indicator in the correction result is scored using preset rules to obtain a rule score corresponding to each anomaly indicator. Each anomaly indicator is then input into a preset anomaly detection model to obtain a model score corresponding to each anomaly indicator. The rule scores and model scores corresponding to each anomaly indicator are then weighted and fused to obtain a fusion score corresponding to each anomaly indicator. Based on the fusion score, anomaly detection results for the current user's multiple accounts are generated.

[0006] This approach collects electricity consumption data from authorized accounts across a centralized electricity account system, providing a data foundation for unified analysis of multi-account electricity consumption data for the current user. This improves the efficiency of unified anomaly detection for multi-account electricity consumption data. Aligning the electricity consumption data with different standards yields several unified-standard electricity consumption records, resolving incomparability issues caused by differences in field naming, units of measurement, and sub-item structures among different electricity accounts. This provides a unified analytical standard for multi-account data, laying the foundation for unified cross-account comparative analysis. Generating corresponding tag sets based on these unified-standard electricity consumption records records contextual information, providing a basis for corrections during subsequent anomaly detection and improving the accuracy of unified anomaly judgment. Indicator analysis of the unified-standard electricity consumption records yields indicator sets, transforming raw electricity consumption data into comparable structured indicators, providing a multi-dimensional analytical foundation for unified anomaly detection. Initial anomaly analysis of each indicator set yields initial anomaly information, quickly filtering out suspicious indicators deviating from historical normal levels, improving the overall efficiency of unified anomaly detection. Using preset labels… The system corrects initial anomaly information using correction rules and tag sets, resulting in corrected results. This eliminates false anomalies caused by normal reasons, reduces the false alarm rate, and ensures the accuracy of unified anomaly detection. Preset rules are used to determine the rules for each anomaly indicator in the corrected results, yielding rule scores for each indicator. This allows for quantified scoring of anomaly indicators based on business experience, ensuring the interpretability of unified anomaly judgment. Each anomaly indicator is input into a preset anomaly detection model, resulting in model scores. This allows for quantified scoring of anomaly indicators based on statistical data, identifying complex anomaly patterns that rules cannot cover, further ensuring the effectiveness of unified anomaly detection. The rule scores and model scores for each anomaly indicator are weighted and fused to obtain a fusion score, combining the advantages of business experience and data patterns to improve the accuracy of unified anomaly judgment. Based on the fusion score, anomaly detection results for multiple accounts of the current user are generated, enabling unified anomaly detection for multiple accounts' electricity consumption data even when the data sources and structures differ. This application solves the problem of difficulty in performing unified anomaly detection for multiple accounts' electricity consumption data when the data sources and structures differ.

[0007] Furthermore, the initial anomaly analysis of each of the aforementioned indicator sets to obtain initial anomaly information specifically includes: Calculate the mean and standard deviation of each basic indicator item in the indicator set; Based on each of the basic indicator items, each of the means and the standard deviation, the initial anomaly score corresponding to each of the basic indicator items is calculated; If the initial anomaly score meets the preset initial anomaly determination conditions, then initial anomaly information is generated based on the basic index item corresponding to the initial anomaly score.

[0008] By calculating the mean and standard deviation of each basic indicator in the indicator set, the historical normal fluctuation range of each indicator can be quantified. Based on the basic indicator and its mean and standard deviation, the initial anomaly score can be calculated, which can quickly screen out suspicious indicators that deviate from the historical normal level. The indicators that meet the preset judgment conditions will generate initial anomaly information, thereby ensuring screening efficiency while providing input for subsequent correction and improving the overall efficiency of unified anomaly detection.

[0009] Furthermore, the process of aligning the electricity consumption data to obtain several unified electricity consumption records specifically includes: By performing a unified field mapping on each of the aforementioned electricity consumption data, several first electricity consumption records are obtained; Perform unit conversion mapping on the first electricity consumption record to obtain several second electricity consumption records; The second electricity consumption record is mapped by a sub-item structure to obtain several electricity consumption records with the same standard.

[0010] By aligning the various electricity consumption data in this way, we can obtain several electricity consumption records with unified standards. This can solve the incomparability problem caused by differences in field naming, measurement units, and sub-item structures among different electricity consumption accounts, and enable multi-account data to have a unified analytical standard, laying the foundation for subsequent unified comparative analysis across accounts.

[0011] Furthermore, the generation of corresponding tag sets and indicator sets based on the unified electricity consumption records specifically includes: The electricity consumption data for the current billing period, the electricity consumption data for the previous billing period, the electricity consumption data for the same billing period in the previous year, the total electricity consumption data for the billing period, the electricity consumption data for peak hours, the electricity consumption data for off-peak hours, the cumulative electricity consumption data for the period, and several sub-items of electricity consumption data in each of the unified standard electricity consumption records are analyzed to obtain several sets of indicators. Based on the start and end dates of the billing period, the billing period coverage time, the peak and valley time configuration, and the cumulative electricity consumption over the period in each of the unified electricity consumption records, several tag sets are generated.

[0012] By analyzing various electricity consumption data in the unified electricity consumption records, the raw electricity consumption data can be transformed into comparable structured indicators, providing a multi-dimensional analytical basis for unified anomaly detection. At the same time, based on information such as the start and end time of the billing period, the billing period coverage time, and the peak and valley time configuration, a tag set can be generated to record contextual information such as cycle type, seasonal holidays, and tiered thresholds, providing a basis for correction in subsequent anomaly detection.

[0013] Furthermore, the analysis of indicators in each of the unified electricity consumption records—including current billing period electricity consumption, previous billing period electricity consumption, same billing period electricity consumption in the previous year, total billing period electricity consumption, peak period electricity consumption, off-peak period electricity consumption, cumulative period electricity consumption, and several sub-items of electricity consumption—results in several sets of indicators, specifically including: For each of the unified electricity consumption records, the month-on-month index value is calculated based on the electricity consumption of the current billing period and the electricity consumption of the previous billing period in the unified electricity consumption records; Based on the current electricity consumption during the current payment period and the electricity consumption during the same payment period in the previous year, the year-on-year indicator value is calculated. Based on the electricity consumption of each item and the total electricity consumption during the billing period, the percentage value of each item is calculated. Based on the peak electricity consumption and the valley electricity consumption, the peak-valley ratio is calculated. Based on the cumulative electricity consumption over the period and the preset tier threshold, the tier progress index value is calculated. The set of indicators is obtained based on the month-on-month indicator value, the year-on-year indicator value, the sub-item percentage indicator value, the peak-to-valley ratio indicator value, and the step progress indicator value.

[0014] By calculating month-on-month, year-on-year, sub-item proportion, peak-to-valley ratio, and tiered progress index values ​​for each unified electricity consumption record, the raw electricity consumption data can be transformed into multi-dimensional and comparable structured indicators, providing a comprehensive analytical basis for unified anomaly detection.

[0015] Furthermore, based on the start and end dates of the billing period, the billing period coverage time, the peak and off-peak time configuration, and the cumulative electricity consumption over the period in each of the unified electricity consumption records, several tag sets are generated, specifically including: For each of the unified electricity consumption records, the start and end times of the billing period in the unified electricity consumption records are analyzed to obtain period type labels; Analyzing the payment period coverage time, seasonal tags and holiday tags are obtained; The cumulative electricity consumption over the period is compared with the threshold value of the tier to obtain the tier critical label; Version identification is performed on the peak and valley time period configuration to obtain peak and valley strategy tags; The tag set is determined based on the cycle type tag, the season tag, the holiday tag, the step threshold tag, and the peak-valley strategy tag.

[0016] By analyzing the start and end dates of billing periods, the period covered by the billing period, the cumulative electricity consumption over the period, and the peak and valley time configurations for each unified electricity consumption record, labels such as period type, season, holiday, tiered threshold, and peak and valley strategy can be obtained. These labels record the contextual information of the electricity consumption data, providing a basis for correction in subsequent anomaly detection. This can effectively eliminate false anomalies caused by normal reasons such as cross-month settlement, holidays, and tiered thresholds, reduce the false alarm rate, and ensure the feasibility of unified anomaly detection.

[0017] Furthermore, the process of constructing the electricity account set specifically includes: Obtain the current user's identity and authorization information; The identity information is used to verify the current user's identity, and the verification result is obtained. If the verification result is successful, the identity information is used to perform association matching in a preset user information database to obtain several candidate accounts corresponding to the current user. The candidate accounts are filtered using the authorization information to obtain a number of authorized accounts; Based on each of the authorized accounts, the electricity account set is generated.

[0018] By using identity information to verify the current user's identity, the legitimacy of the user's identity can be ensured. After the verification is successful, candidate accounts are matched based on the identity information, and then filtered in combination with authorization information to ensure that only accounts that the current user has the right to access are filtered out. Based on the filtered authorized accounts, a set of electricity accounts is generated, which realizes the association of discrete electricity accounts to the same user, laying a data foundation for subsequent unified analysis of multiple accounts.

[0019] Secondly, an embodiment of the present invention provides an anomaly detection device for multi-account electricity consumption data, including a first module, a second module and a third module; The first module is used to collect electricity consumption data of each authorized account in the electricity account set, wherein the electricity account set includes all the authorized accounts of the current user; The second module is used to perform caliber alignment processing on each of the electricity consumption data to obtain several unified caliber electricity consumption records, and to generate corresponding tag sets and indicator sets based on each of the unified caliber electricity consumption records. The third module is used to perform initial anomaly analysis on each set of indicators to obtain initial anomaly information, correct the initial anomaly information using each set of tags to obtain a correction result, score each anomaly indicator item in the correction result using preset rules to obtain a rule score corresponding to each anomaly indicator item, input each anomaly indicator item into a preset anomaly detection model to obtain a model score corresponding to each anomaly indicator item, perform weighted fusion of the rule score and model score corresponding to each anomaly indicator item to obtain a fusion score corresponding to each anomaly indicator item, and generate anomaly detection results for multiple accounts of the current user based on the fusion score.

[0020] The first module collects electricity consumption data from authorized accounts in a centralized electricity account database, providing a data foundation for unified analysis of multi-account electricity consumption data for the current user, thereby improving the efficiency of unified anomaly detection for multi-account electricity consumption data. The second module aligns the electricity consumption data, resulting in several unified-caliber electricity consumption records. This resolves the incomparability issues caused by differences in field naming, measurement units, and sub-item structures among different electricity accounts, ensuring a unified analytical caliber for multi-account data and laying the foundation for unified cross-account comparative analysis. Based on each unified-caliber electricity consumption record, a corresponding tag set is generated, recording contextual information and providing a basis for corrections in subsequent anomaly detection, improving the accuracy of unified anomaly judgment. Indicator analysis is performed on the unified-caliber electricity consumption records to obtain indicator sets, transforming raw electricity consumption data into comparable structured indicators, providing a multi-dimensional analytical foundation for unified anomaly detection. The third module performs initial anomaly analysis on each indicator set, obtaining initial anomaly information, which can quickly filter out suspicious indicators deviating from historical normal levels, improving the overall efficiency of unified anomaly detection. The system corrects initial anomaly information using preset label correction rules and various label sets, eliminating false anomalies caused by normal reasons, reducing the false alarm rate, and ensuring the accuracy of unified anomaly detection. It then uses preset rules to determine the rules for each anomaly indicator in the correction results, obtaining a rule score for each indicator. This allows for quantitative scoring of anomaly indicators based on business experience, ensuring the interpretability of unified anomaly judgment. Finally, it inputs each anomaly indicator into a preset anomaly detection model, obtaining a model score for each indicator. This allows for quantitative scoring of anomaly indicators based on statistical data, identifying complex anomaly patterns that rules cannot cover, further ensuring the effectiveness of unified anomaly detection. The system then weights and fuses the rule scores and model scores for each anomaly indicator to obtain a fusion score, combining the advantages of business experience and data patterns to improve the accuracy of unified anomaly judgment. Based on the fusion score, it generates anomaly detection results for multiple accounts of the current user, enabling unified anomaly detection for multiple accounts even when the sources and structures of electricity consumption data differ.

[0021] Thirdly, another embodiment of the present invention provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device or apparatus where the computer-readable storage medium is located to perform an anomaly detection method for multi-account electricity consumption data.

[0022] Fourthly, another embodiment of the present invention provides a computer program product, including a computer program or instructions, which, when executed by a communication device, implements a method for detecting abnormal electricity consumption data of multiple accounts. Attached Figure Description

[0023] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating an embodiment of the anomaly detection method for multi-account electricity consumption data provided in this application; Figure 2 This is a flowchart illustrating steps S201 to S202 provided in this application; Figure 3 This is a schematic diagram of the structure of an anomaly detection device for multi-account electricity consumption data provided in this application. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0027] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0028] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0029] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0030] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0031] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this application according to the specific circumstances.

[0032] In the field of power data processing technology, unified anomaly detection of multi-account electricity consumption data can help users optimize electricity consumption strategies and prevent electricity consumption risks. Existing methods use time-series-based curve comparison for anomaly detection, but these methods have fundamental limitations: firstly, they only analyze single electricity accounts and cannot handle cross-account comparative analysis and unified anomaly detection in multi-account scenarios; secondly, electricity consumption data from different accounts differ in field naming, units of measurement, and sub-item structures, and existing technologies lack a unified processing mechanism for these data source and structural differences. This makes direct comparison and analysis of multi-account data impossible, and the anomaly detection results fail to accurately reflect the true electricity consumption situation in multi-account scenarios.

[0033] See Figure 1 In order to solve the problem that it is difficult to perform unified anomaly detection on electricity consumption data of multiple accounts when there are differences in the source and structure of electricity consumption data of multiple accounts, an embodiment of the present invention provides an anomaly detection method for electricity consumption data of multiple accounts, including steps S101 to S103. Step S101: Collect electricity consumption data of each authorized account in the electricity account set, wherein the electricity account set includes all the authorized accounts of the current user; In some embodiments, electricity consumption data of each electricity account in the electricity account set is collected. The electricity account set includes all the authorized accounts of the current user. Specifically, this includes: reading the account identifier list in the completed electricity account set; according to each account identifier in the account identifier list, pulling the electricity consumption data corresponding to each authorized account from the power data source, including the electricity consumption, item details and billing information of each account; and organizing the pulled electricity consumption data according to the account identifier to form an electricity data set corresponding to the electricity account set.

[0034] In some embodiments, the process of constructing the electricity account set specifically includes: obtaining the identity information and authorization information of the current user; verifying the identity of the current user using the identity information to obtain a verification result; if the verification result is successful, performing association matching in a preset user information database based on the identity information to obtain several candidate accounts corresponding to the current user; filtering each candidate account using the authorization information to obtain several authorized accounts; and generating the electricity account set based on each authorized account. Specifically, the system receives power service requests initiated by users when they actively handle power business through the power company's online service channels (including mobile applications or online business halls). It extracts the user's identity and authorization information from the request message. The identity information is used to verify the user's login status and role type, validating the user's identity. After successful verification, the system performs correlation matching in a pre-set user information database based on the identity information, querying all electricity accounts corresponding to that identity to obtain a candidate account set. The system parses the account number range, shared member list, and validity period from the authorization information, using the parsed account number range to filter the candidate account set, retaining accounts with account numbers within the authorization range to obtain authorizable accounts. For authorizable accounts belonging to the same transformer area or the same master-slave topology, the system aggregates and labels the account numbers, recording the topological relationship between the master table and the sub-tables. Based on the aggregated and labeled authorizable accounts, shared member list, and validity period, a power account set is generated, and a unique account set identifier is assigned to each power account set.

[0035] It should be noted that the permission boundaries are bound to the electricity account set, and at least the range of viewable account numbers, shareable objects, exportable fields and validity period are limited. If the authorization information is missing, expired or withdrawn, an empty electricity account set will be output and a rejection reason code will be returned, prompting the user to re-authorize on the interface.

[0036] By using identity information to verify the current user's identity, the legitimacy of the user's identity can be ensured. After the verification is successful, candidate accounts are matched based on the identity information, and then filtered in combination with authorization information to ensure that only accounts that the current user has the right to access are filtered out. Based on the filtered authorized accounts, a set of electricity accounts is generated, which realizes the association of discrete electricity accounts to the same user, laying a data foundation for subsequent unified analysis of multiple accounts.

[0037] Step S102: Perform caliber alignment processing on each of the electricity consumption data to obtain several unified caliber electricity consumption records, and generate corresponding tag sets and indicator sets based on each of the unified caliber electricity consumption records. In some embodiments, the step of aligning the electricity consumption data to obtain several unified-caliber electricity consumption records specifically includes: performing unified field mapping on the electricity consumption data to obtain several first electricity consumption records; performing unit conversion mapping on the first electricity consumption records to obtain several second electricity consumption records; and performing itemized structure mapping on the second electricity consumption records to obtain several unified-caliber electricity consumption records. Specifically, according to the preset unified field mapping rules, the field names in each electricity consumption data are mapped to standard field names. For example, the total electricity consumption is mapped to the total power consumption, resulting in the first electricity consumption record. According to the preset unit conversion mapping rules, the power consumption units in the first electricity consumption record are uniformly converted. For example, all power consumption values ​​are uniformly converted to kilowatt-hours, the multiplier is converted, the precision is normalized, and negative values ​​are marked as positive, resulting in the second electricity consumption record. According to the preset sub-item structure mapping rules, the tiered sub-items, peak-valley sub-items, basic electricity fee sub-items, power regulation electricity fee sub-items, and demand electricity fee sub-items in the second electricity consumption record are merged into a unified sub-item hierarchical structure to construct a sub-item detailed dictionary, resulting in a unified standard electricity consumption record. The original field path is retained for each standard field, and the source field, conversion formula, and multiplier source are recorded.

[0038] It should be noted that the caliber alignment process includes three steps: unified field mapping, unit conversion mapping, and itemized structure mapping. This can solve the incomparability problem caused by differences in field naming, measurement units, and itemized structures among different electricity accounts, and enable multi-account data to have a unified analytical caliber. If the electricity consumption data of a certain account for a certain payment period is missing, the missing data will be filled in according to the historical window (such as backfilling with the most recent valid payment period and marking the missing data). At the same time, the data quality reason code will be output, and the missing data filling action will be written into the evidence chain.

[0039] By aligning the various electricity consumption data in this way, we can obtain several electricity consumption records with unified standards. This can solve the incomparability problem caused by differences in field naming, measurement units, and sub-item structures among different electricity consumption accounts, and enable multi-account data to have a unified analytical standard, laying the foundation for subsequent unified comparative analysis across accounts.

[0040] See Figure 2 In some embodiments, the step of generating corresponding tag sets and indicator sets based on each of the unified standard electricity consumption records specifically includes steps S201 to S202. Step S201: Perform index analysis on the current billing period electricity consumption, previous billing period electricity consumption, previous year same billing period electricity consumption, total billing period electricity consumption, peak period electricity consumption, valley period electricity consumption, cumulative period electricity consumption, and several sub-item electricity consumption in each of the unified standard electricity consumption records to obtain several sets of the aforementioned indicators; In some embodiments, the step of analyzing the electricity consumption for the current billing period, the electricity consumption for the previous billing period, the electricity consumption for the same billing period of the previous year, the total electricity consumption for the billing period, the peak electricity consumption, the valley electricity consumption, the cumulative electricity consumption over a period, and several sub-items of electricity consumption in each of the unified-caliber electricity consumption records to obtain several sets of indicators specifically includes: for each of the unified-caliber electricity consumption records, calculating a month-on-month indicator value based on the electricity consumption for the current billing period and the electricity consumption for the previous billing period; calculating a year-on-year indicator value based on the electricity consumption for the current billing period and the electricity consumption for the same billing period of the previous year; calculating a sub-item percentage indicator value based on each sub-item of electricity consumption and the total electricity consumption for the billing period; calculating a peak-valley ratio indicator value based on the electricity consumption for the peak period and the electricity consumption for the valley period; calculating a tiered progress indicator value based on the cumulative electricity consumption over a period and a preset tiered threshold; and obtaining the indicator set based on the month-on-month indicator value, the year-on-year indicator value, the sub-item percentage indicator value, the peak-valley ratio indicator value, and the tiered progress indicator value. Specifically, for each unified electricity consumption record, a month-on-month indicator value is calculated based on the current and previous payment period's electricity consumption. The month-on-month indicator value is calculated by subtracting the previous payment period's electricity consumption from the current payment period's electricity consumption, and then dividing the difference by the larger of the previous payment period's electricity consumption and a minimum constant. A year-on-year indicator value is calculated based on the current payment period's electricity consumption and the same payment period of the previous year's electricity consumption. The year-on-year indicator value is calculated by subtracting the previous year's electricity consumption from the current payment period's electricity consumption, and then dividing the difference by the larger of the previous year's electricity consumption and a minimum constant. Finally, based on each item's electricity consumption and the total payment period's electricity consumption, item-by-item indicators are calculated. The percentage indicators are calculated as follows: the percentage of each item's electricity consumption is divided by the larger of the total electricity consumption over the billing period and a minimum constant; the peak-to-valley ratio is calculated based on peak and valley electricity consumption, and is calculated by dividing peak electricity consumption by valley electricity consumption and a minimum constant; the tiered progress indicator is calculated based on the cumulative electricity consumption over the period and a preset tiered threshold, and is calculated by dividing the cumulative electricity consumption over the period by the tiered threshold; and the set of indicators is obtained based on the month-on-month indicator, year-on-year indicator, percentage of each item's electricity consumption, peak-to-valley ratio, and tiered progress indicator.

[0041] In some embodiments, the analysis of the current billing period electricity consumption, the previous billing period electricity consumption, the same billing period electricity consumption of the previous year, the total billing period electricity consumption, peak period electricity consumption, valley period electricity consumption, cumulative period electricity consumption, and several sub-items of electricity consumption in each of the unified caliber electricity consumption records, to obtain relevant formulas for several sets of indicators, specifically includes: Formula for calculating month-on-month index values: ; Formula for calculating year-on-year indicator values: ; Formula for calculating the percentage value of each item: ; Formula for calculating peak-to-valley ratio: ; Formula for calculating the step progress index value: ; In the formula, This indicates the electricity consumption during the current billing period; This indicates the electricity consumption in the previous billing period. This indicates the electricity consumption during the same accounting period in the previous year; To prevent constants with a denominator of zero; This represents the electricity consumption of the i-th item; This indicates the total electricity consumption over the entire billing period; This indicates peak electricity consumption. Indicates electricity consumption during off-peak hours; This indicates the cumulative electricity consumption over the period. This indicates the preset threshold for each step.

[0042] By calculating month-on-month, year-on-year, sub-item proportion, peak-to-valley ratio, and tiered progress index values ​​for each unified electricity consumption record, the raw electricity consumption data can be transformed into multi-dimensional and comparable structured indicators, providing a comprehensive analytical basis for unified anomaly detection.

[0043] Step S202: Based on the start and end dates of the billing period, the billing period coverage time, the peak and valley time configuration, and the cumulative electricity consumption of the cycle in each of the unified caliber electricity consumption records, generate several tag sets.

[0044] In some embodiments, generating several tag sets based on the billing period start and end times, billing period coverage time periods, peak and valley time period configurations, and the cumulative electricity consumption over a period in each of the unified-caliber electricity consumption records specifically includes: analyzing the billing period start and end times in each of the unified-caliber electricity consumption records to obtain a period type tag; analyzing the billing period coverage time periods to obtain a seasonal tag and a holiday tag; comparing the cumulative electricity consumption over a period with the tiered threshold to obtain a tiered threshold tag; identifying the version of the peak and valley time period configurations to obtain a peak and valley strategy tag; and determining the tag set based on the period type tag, the seasonal tag, the holiday tag, the tiered threshold tag, and the peak and valley strategy tag. Specifically, for each unified electricity consumption record, the start and end times of the billing period are analyzed to obtain a cycle type label, which identifies whether the billing period belongs to a calendar month, a cross-month settlement, or a supplementary meter reading settlement. The time period covered by the billing period in the unified electricity consumption record is analyzed to obtain a season label and a holiday label. The season label identifies the season to which the billing period belongs, and the holiday label identifies whether the billing period includes statutory holidays. The cumulative electricity consumption of the cycle in the unified electricity consumption record is compared with the tiered threshold to obtain a tiered threshold label, which identifies whether the current cumulative electricity consumption of the cycle is close to or exceeds the tiered threshold. The peak-valley time period configuration in the unified electricity consumption record is version identified to obtain a peak-valley strategy label, which identifies the currently used peak-valley strategy version. Based on the cycle type label, season label, holiday label, tiered threshold label, and peak-valley strategy label, a label set is determined.

[0045] By analyzing the start and end dates of billing periods, the period covered by the billing period, the cumulative electricity consumption over the period, and the peak and valley time configurations for each unified electricity consumption record, labels such as period type, season, holiday, tiered threshold, and peak and valley strategy can be obtained. These labels record the contextual information of the electricity consumption data, providing a basis for correction in subsequent anomaly detection. This can effectively eliminate false anomalies caused by normal reasons such as cross-month settlement, holidays, and tiered thresholds, reduce the false alarm rate, and ensure the feasibility of unified anomaly detection.

[0046] By analyzing various electricity consumption data in the unified electricity consumption records, the raw electricity consumption data can be transformed into comparable structured indicators, providing a multi-dimensional analytical basis for unified anomaly detection. At the same time, based on information such as the start and end time of the billing period, the billing period coverage time, and the peak and valley time configuration, a tag set can be generated to record contextual information such as cycle type, seasonal holidays, and tiered thresholds, providing a basis for correction in subsequent anomaly detection.

[0047] Step S103: Perform initial anomaly analysis on each set of indicators to obtain initial anomaly information; correct the initial anomaly information using each set of tags to obtain correction results; score each anomaly indicator item in the correction results using preset rules to obtain rule scores corresponding to each anomaly indicator item; input each anomaly indicator item into a preset anomaly detection model to obtain model scores corresponding to each anomaly indicator item; perform weighted fusion of the rule scores and model scores corresponding to each anomaly indicator item to obtain fusion scores corresponding to each anomaly indicator item; and generate anomaly detection results for multiple accounts of the current user based on the fusion scores. In some embodiments, the initial anomaly analysis of each set of indicators to obtain initial anomaly information specifically includes: calculating the mean and standard deviation of each basic indicator item in the set of indicators; calculating the initial anomaly score corresponding to each basic indicator item based on each basic indicator item, each mean and the standard deviation; and generating initial anomaly information based on the basic indicator item corresponding to the initial anomaly score if the initial anomaly score meets a preset initial anomaly judgment condition. Specifically, for each indicator set, the mean and standard deviation of each basic indicator item in the indicator set (i.e., the month-on-month indicator value, year-on-year indicator value, sub-item percentage indicator value, peak-to-valley ratio indicator value, and step progress indicator value mentioned above) are calculated. The mean is calculated by adding the values ​​of each basic indicator item within a specified window and dividing by the window length. The standard deviation is calculated by summing the squares of the differences between each basic indicator item value and the mean, dividing by the window length, and then taking the square root. Based on each basic indicator item, each mean, and the standard deviation, the initial anomaly score corresponding to each basic indicator item is calculated. The initial anomaly score is calculated by subtracting the mean from the basic indicator item and then dividing by the standard deviation. If the absolute value of the initial anomaly score is greater than or equal to a preset anomaly threshold, then initial anomaly information is generated based on the basic indicator item corresponding to the initial anomaly score.

[0048] It should be noted that the historical window length in the initial anomaly score calculation is configurable (e.g., the last 6 or 12 periods), and the threshold can be adjusted according to the anomaly sensitivity level in the user profile. If the standard deviation is too small or the window is insufficient, causing the standard score to be unstable, switch to robust statistics or only output trend prompts without alarms, and record the downgrade strategy in the evidence chain.

[0049] In some embodiments, the formula for performing initial anomaly analysis on each of the indicator sets to obtain initial anomaly information specifically includes: Initial anomaly score: ; In the formula, Indicates basic indicator items; This represents the average value within a historical window. This represents the standard deviation within the historical window.

[0050] By calculating the mean and standard deviation of each basic indicator in the indicator set, the historical normal fluctuation range of each indicator can be quantified. Based on the basic indicator and its mean and standard deviation, the initial anomaly score can be calculated, which can quickly screen out suspicious indicators that deviate from the historical normal level. The indicators that meet the preset judgment conditions will generate initial anomaly information, thereby ensuring screening efficiency while providing input for subsequent correction and improving the overall efficiency of unified anomaly detection.

[0051] In some embodiments, the initial anomaly information is corrected using each of the tag sets to obtain a correction result. Specifically, this includes: determining the unified standard electricity consumption record corresponding to each initial anomaly in the initial anomaly information; based on the corresponding tag set, correcting the initial anomaly according to the period type tag, seasonal tag, holiday tag, and step-critical tag in the tag set; if the period type tag is a cross-month settlement or supplementary meter reading settlement, then the weight of growth rate anomalies is reduced or the judgment is delayed; if the seasonal tag or holiday tag indicates seasonal fluctuations or holiday impacts, then the seasonal baseline is used to replace the historical average for re-judgment; if the step-critical tag is a critical state, then the judgment threshold for step-progress anomalies is increased; and correcting each initial anomaly in the initial anomaly information according to these correction strategies to obtain a correction result.

[0052] In some embodiments, each abnormal indicator item in the correction result is scored using preset rules to obtain a rule score corresponding to each abnormal indicator item. Specifically, this includes: determining whether each abnormal indicator item in the correction result triggers a preset rule according to a preset rule configuration file. The preset rules include rules for sudden changes in the proportion of sub-items, rules for tiered thresholds, rules for changes in peak-to-valley ratios, rules for overdue payment risk, and rules for differences in billing cycles. Each rule corresponds to a weight value. When an abnormal indicator item meets the triggering condition of a certain rule, the weight of that rule is added to the rule score. After traversing all rules, the rule score corresponding to that abnormal indicator item is obtained.

[0053] To further explain, the trigger condition for the sub-item percentage mutation rule is that the absolute value of the sub-item percentage change rate is greater than or equal to the sub-item percentage change threshold, and the weight is the first weight value; the trigger condition for the step critical rule is that the step critical label is in a critical state and the absolute value of one minus the step progress is less than or equal to the step critical threshold, and the weight is the second weight value; the trigger condition for the peak-valley ratio change rule is that the absolute value of the peak-valley ratio change rate is greater than or equal to the peak-valley ratio change threshold and the peak-valley strategy label has changed, and the weight is the third weight value; if an abnormal indicator item triggers both the sub-item percentage mutation rule and the step critical rule, but does not trigger the peak-valley ratio change rule, then the rule score of the abnormal indicator item is the sum of the first weight value and the second weight value. If all three rules are triggered simultaneously, then the rule score is the sum of the first weight value, the second weight value, and the third weight value.

[0054] In some embodiments, each of the abnormal indicator items is input into a preset anomaly detection model to obtain a model score corresponding to each of the abnormal indicator items. Specifically, this includes: extracting feature data for each abnormal indicator item, including current indicator value, historical mean, historical standard deviation, initial anomaly score, month-on-month indicator value, year-on-year indicator value, sub-item percentage indicator value, step progress indicator value, peak-to-valley ratio indicator value, cycle type code, seasonal code, holiday identifier, and peak-to-valley strategy code. The feature data is input into the preset anomaly detection model, and the anomaly detection model outputs a probability value between 0 and 1. This probability value is the model score, which indicates the probability that the abnormal indicator item is abnormal.

[0055] It should be noted that the period type encoding can be represented by numerical codes, such as 0 for natural month, 1 for cross-month settlement, and 2 for supplementary meter reading settlement. The seasonal encoding can also be represented by numerical codes, such as 0 for spring, 1 for summer, 2 for autumn, and 3 for winter. The holiday identifier can be represented by Boolean values, such as 1 for including holidays and 0 for not including holidays. The peak-valley strategy encoding can also be represented by numerical codes, such as 0 for version A and 1 for version B. The anomaly detection model can use a logistic regression model, a gradient boosting tree model, or a neural network model. When training the anomaly detection model, historical electricity consumption indicators are used as input features, and historical anomaly labeling results are used as supervision labels. The model parameters are obtained by training by minimizing the cross-entropy loss function.

[0056] In some embodiments, the rule scores and model scores corresponding to each of the abnormal indicator items are weighted and fused to obtain a fusion score corresponding to each of the abnormal indicator items. Specifically, this includes: weighting and fusing the rule scores and model scores corresponding to each of the abnormal indicator items according to a preset fusion weight to obtain a fusion score.

[0057] In some embodiments, a weighted fusion of the rule score and model score corresponding to each of the anomaly indicators is performed to obtain a relevant formula for the fusion score corresponding to each of the anomaly indicators, specifically including: Formula for calculating the fusion score: ; In the formula, Represents the rule score; Indicates the model score; This represents the fusion weight, with a value ranging from 0 to 1.

[0058] In some embodiments, generating anomaly detection results for multiple accounts of the current user based on the fusion score specifically includes: comparing the fusion score with a second preset threshold; if the fusion score is greater than or equal to the second preset threshold, confirming the anomaly indicator as an anomaly; otherwise, treating it as a weak anomaly and only providing a notification without issuing an alarm. The anomaly detection results for multiple accounts of the current user are generated based on the confirmed anomaly indicator, and the anomaly detection results are output.

[0059] It should be noted that the anomaly detection results include the cause codes corresponding to the anomalies. The cause codes are matched from a preset cause code dictionary, which is used to uniformly manage the encoding, name, triggering conditions, evidence field set, recommended explanation templates, and handling suggestions of the cause codes. The cause codes include cause codes for sudden changes in sub-item proportions, tiered criticality cause codes, peak-to-valley ratio changes, overdue or supplementary payment cause codes, billing cycle differences cause codes, total and sub-table balance deviation cause codes, line loss anomaly cause codes, and demand exceeding limits cause codes. Each cause code corresponds to a severity level, including alert level, warning level, and severe level, which are used to display different alarm styles in the visualization interface.

[0060] For example, after generating the anomaly detection results, a visualization step is also included: the system generates a view specification object based on the user role, terminal size, and task scenario. The view specification object defines the component set, scale rules, information density threshold, and default sorting. The interface is rendered based on the view specification object and the interaction state object to generate the user interface, and key view data is written to the rendering cache. When the user clicks on the anomaly card, the interaction state machine focuses on the anomaly period and indicator, highlights the anomaly point in the chart, and opens the evidence drawer to display the evidence chain.

[0061] For example, after generating the anomaly detection results, the system also includes a behavior feedback and optimization step: the system collects user interaction behavior with the anomaly detection results and generates behavior logs, which include interaction events such as clicks, switches, drill-downs, confirmations, or sharing; the system calculates performance evaluation metrics based on the behavior logs, which include task completion rate, dwell time, number of rollbacks, and anomaly explanation click rate; based on the performance evaluation metrics, the system generates an update package, which includes threshold changes, rule changes, model changes, linkage changes, and explanation text changes. The update package is used to optimize the preset rules, preset anomaly detection models, preset thresholds, and cause code dictionaries, forming a closed-loop iteration.

[0062] This approach collects electricity consumption data from authorized accounts across a centralized electricity account system, providing a data foundation for unified analysis of multi-account electricity consumption data for the current user. This improves the efficiency of unified anomaly detection for multi-account electricity consumption data. Aligning the electricity consumption data with different standards yields several unified-standard electricity consumption records, resolving incomparability issues caused by differences in field naming, units of measurement, and sub-item structures among different electricity accounts. This provides a unified analytical standard for multi-account data, laying the foundation for unified cross-account comparative analysis. Generating corresponding tag sets based on these unified-standard electricity consumption records records contextual information, providing a basis for corrections during subsequent anomaly detection and improving the accuracy of unified anomaly judgment. Indicator analysis of the unified-standard electricity consumption records yields indicator sets, transforming raw electricity consumption data into comparable structured indicators, providing a multi-dimensional analytical foundation for unified anomaly detection. Initial anomaly analysis of each indicator set yields initial anomaly information, quickly filtering out suspicious indicators deviating from historical normal levels, improving the overall efficiency of unified anomaly detection. Using preset labels… The system corrects initial anomaly information using correction rules and tag sets, resulting in corrected results. This eliminates false anomalies caused by normal reasons, reduces the false alarm rate, and ensures the accuracy of unified anomaly detection. Preset rules are used to determine the rules for each anomaly indicator in the corrected results, yielding rule scores for each indicator. This allows for quantified scoring of anomaly indicators based on business experience, ensuring the interpretability of unified anomaly judgment. Each anomaly indicator is input into a preset anomaly detection model, resulting in model scores. This allows for quantified scoring of anomaly indicators based on statistical data, identifying complex anomaly patterns that rules cannot cover, further ensuring the effectiveness of unified anomaly detection. The rule scores and model scores for each anomaly indicator are weighted and fused to obtain a fusion score, combining the advantages of business experience and data patterns to improve the accuracy of unified anomaly judgment. Based on the fusion score, anomaly detection results for multiple accounts of the current user are generated, enabling unified anomaly detection for multiple accounts' electricity consumption data even when the data sources and structures differ. This application solves the problem of difficulty in performing unified anomaly detection for multiple accounts' electricity consumption data when the data sources and structures differ.

[0063] See Figure 3 Based on the above method embodiments, corresponding device embodiments are provided; One embodiment of the present invention provides an anomaly detection device for multi-account electricity consumption data, including a first module 100, a second module 200 and a third module 300; The first module 100 is used to collect electricity consumption data of each authorized account in the electricity account set, wherein the electricity account set includes all the authorized accounts of the current user; The second module 200 is used to perform caliber alignment processing on each of the electricity consumption data to obtain several unified caliber electricity consumption records, and to generate a corresponding tag set and indicator set based on each of the unified caliber electricity consumption records. The third module 300 is used to perform initial anomaly analysis on each set of indicators to obtain initial anomaly information, correct the initial anomaly information using each set of tags to obtain a correction result, score each anomaly indicator item in the correction result using preset rules to obtain a rule score corresponding to each anomaly indicator item, input each anomaly indicator item into a preset anomaly detection model to obtain a model score corresponding to each anomaly indicator item, perform weighted fusion of the rule score and model score corresponding to each anomaly indicator item to obtain a fusion score corresponding to each anomaly indicator item, and generate anomaly detection results for multiple accounts of the current user based on the fusion score.

[0064] The first module collects electricity consumption data from authorized accounts in a centralized electricity account database, providing a data foundation for unified analysis of multi-account electricity consumption data for the current user, thereby improving the efficiency of unified anomaly detection for multi-account electricity consumption data. The second module aligns the electricity consumption data, resulting in several unified-caliber electricity consumption records. This resolves the incomparability issues caused by differences in field naming, measurement units, and sub-item structures among different electricity accounts, ensuring a unified analytical caliber for multi-account data and laying the foundation for unified cross-account comparative analysis. Based on each unified-caliber electricity consumption record, a corresponding tag set is generated, recording contextual information and providing a basis for corrections in subsequent anomaly detection, improving the accuracy of unified anomaly judgment. Indicator analysis is performed on the unified-caliber electricity consumption records to obtain indicator sets, transforming raw electricity consumption data into comparable structured indicators, providing a multi-dimensional analytical foundation for unified anomaly detection. The third module performs initial anomaly analysis on each indicator set, obtaining initial anomaly information, which can quickly filter out suspicious indicators deviating from historical normal levels, improving the overall efficiency of unified anomaly detection. The system corrects initial anomaly information using preset label correction rules and various label sets, eliminating false anomalies caused by normal reasons, reducing the false alarm rate, and ensuring the accuracy of unified anomaly detection. It then uses preset rules to determine the rules for each anomaly indicator in the correction results, obtaining a rule score for each indicator. This allows for quantitative scoring of anomaly indicators based on business experience, ensuring the interpretability of unified anomaly judgment. Finally, it inputs each anomaly indicator into a preset anomaly detection model, obtaining a model score for each indicator. This allows for quantitative scoring of anomaly indicators based on statistical data, identifying complex anomaly patterns that rules cannot cover, further ensuring the effectiveness of unified anomaly detection. The system then weights and fuses the rule scores and model scores for each anomaly indicator to obtain a fusion score, combining the advantages of business experience and data patterns to improve the accuracy of unified anomaly judgment. Based on the fusion score, it generates anomaly detection results for multiple accounts of the current user, enabling unified anomaly detection for multiple accounts even when the sources and structures of electricity consumption data differ.

[0065] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can implement the method for detecting anomalies in multi-account electricity consumption data provided by any of the above-described method embodiments of the present invention.

[0066] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0067] Based on the above-described embodiment of the method for detecting anomalies in multi-account electricity consumption data, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the method for detecting anomalies in multi-account electricity consumption data according to any embodiment of the present invention.

[0068] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0069] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0070] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0071] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the anomaly detection method for multi-account electricity consumption data described in any of the above-described method embodiments of the present invention.

[0072] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0073] Based on the above-described method embodiments, another embodiment of the present invention provides a computer program product, including a computer program or instructions, which, when executed by a communication device, implements a method for detecting abnormal electricity consumption data of multiple accounts.

[0074] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for detecting anomalies in multi-account electricity consumption data, characterized in that, include: Collect electricity consumption data from each authorized account in the electricity account set, wherein the electricity account set includes all the authorized accounts of the current user; The electricity consumption data are aligned to obtain several unified electricity consumption records, and a corresponding tag set and indicator set are generated based on each unified electricity consumption record. An initial anomaly analysis is performed on each set of indicators to obtain initial anomaly information. The initial anomaly information is then corrected using each set of tags to obtain a correction result. Each anomaly indicator in the correction result is scored using preset rules to obtain a rule score corresponding to each anomaly indicator. Each anomaly indicator is then input into a preset anomaly detection model to obtain a model score corresponding to each anomaly indicator. The rule scores and model scores corresponding to each anomaly indicator are then weighted and fused to obtain a fusion score corresponding to each anomaly indicator. Based on the fusion score, anomaly detection results for the current user's multiple accounts are generated.

2. The method for detecting anomalies in multi-account electricity consumption data as described in claim 1, characterized in that, The initial anomaly analysis of each of the aforementioned indicator sets to obtain initial anomaly information specifically includes: Calculate the mean and standard deviation of each basic indicator item in the indicator set; Based on each of the basic indicator items, each of the means and the standard deviation, the initial anomaly score corresponding to each of the basic indicator items is calculated; If the initial anomaly score meets the preset initial anomaly determination conditions, then initial anomaly information is generated based on the basic index item corresponding to the initial anomaly score.

3. The method for detecting anomalies in multi-account electricity consumption data as described in claim 1, characterized in that, The process of aligning the electricity consumption data to obtain several electricity consumption records with unified standards includes: By performing a unified field mapping on each of the aforementioned electricity consumption data, several first electricity consumption records are obtained; Perform unit conversion mapping on the first electricity consumption record to obtain several second electricity consumption records; The second electricity consumption record is mapped by a sub-item structure to obtain several electricity consumption records with the same standard.

4. The method for detecting anomalies in multi-account electricity consumption data as described in claim 1, characterized in that, The generation of corresponding tag sets and indicator sets based on the unified electricity consumption records specifically includes: The electricity consumption data for the current billing period, the electricity consumption data for the previous billing period, the electricity consumption data for the same billing period in the previous year, the total electricity consumption data for the billing period, the electricity consumption data for peak hours, the electricity consumption data for off-peak hours, the cumulative electricity consumption data for the period, and several sub-items of electricity consumption data in each of the unified standard electricity consumption records are analyzed to obtain several sets of indicators. Based on the start and end dates of the billing period, the billing period coverage time, the peak and valley time configuration, and the cumulative electricity consumption over the period in each of the unified electricity consumption records, several tag sets are generated.

5. The method for detecting anomalies in multi-account electricity consumption data as described in claim 4, characterized in that, The analysis of the electricity consumption data in each of the unified electricity consumption records for the current billing period, the previous billing period, the same billing period of the previous year, the total billing period, peak hour electricity consumption, off-peak hour electricity consumption, cumulative electricity consumption over a period, and several sub-items of electricity consumption yields several sets of indicators, specifically including: For each of the unified electricity consumption records, the month-on-month index value is calculated based on the electricity consumption of the current billing period and the electricity consumption of the previous billing period in the unified electricity consumption records; Based on the current electricity consumption during the current payment period and the electricity consumption during the same payment period in the previous year, the year-on-year indicator value is calculated. Based on the electricity consumption of each item and the total electricity consumption during the billing period, the percentage value of each item is calculated. Based on the peak electricity consumption and the valley electricity consumption, the peak-valley ratio is calculated. Based on the cumulative electricity consumption over the period and the preset tier threshold, the tier progress index value is calculated. The set of indicators is obtained based on the month-on-month indicator value, the year-on-year indicator value, the sub-item percentage indicator value, the peak-to-valley ratio indicator value, and the step progress indicator value.

6. The method for detecting anomalies in multi-account electricity consumption data as described in claim 5, characterized in that, Based on the start and end dates of the billing period, the billing period coverage time, the peak and off-peak time configuration, and the cumulative electricity consumption over the period in each of the unified electricity consumption records, several tag sets are generated, specifically including: For each of the unified electricity consumption records, the start and end times of the billing period in the unified electricity consumption records are analyzed to obtain period type labels; Analyzing the payment period coverage time, seasonal tags and holiday tags are obtained; The cumulative electricity consumption over the period is compared with the threshold value of the tier to obtain the tier critical label; Version identification is performed on the peak and valley time period configuration to obtain peak and valley strategy tags; The tag set is determined based on the cycle type tag, the season tag, the holiday tag, the step threshold tag, and the peak-valley strategy tag.

7. The method for detecting anomalies in multi-account electricity consumption data as described in claim 1, characterized in that, The process of constructing the electricity account set specifically includes: Obtain the current user's identity and authorization information; The identity information is used to verify the current user's identity, and the verification result is obtained. If the verification result is successful, the identity information is used to perform association matching in a preset user information database to obtain several candidate accounts corresponding to the current user. The candidate accounts are filtered using the authorization information to obtain a number of authorized accounts; Based on each of the authorized accounts, the electricity account set is generated.

8. A device for detecting anomalies in multi-account electricity consumption data, characterized in that, It includes Module 1, Module 2, and Module 3; The first module is used to collect electricity consumption data of each authorized account in the electricity account set, wherein the electricity account set includes all the authorized accounts of the current user; The second module is used to perform caliber alignment processing on each of the electricity consumption data to obtain several unified caliber electricity consumption records, and to generate corresponding tag sets and indicator sets based on each of the unified caliber electricity consumption records. The third module is used to perform initial anomaly analysis on each set of indicators to obtain initial anomaly information, correct the initial anomaly information using each set of tags to obtain a correction result, score each anomaly indicator item in the correction result using preset rules to obtain a rule score corresponding to each anomaly indicator item, input each anomaly indicator item into a preset anomaly detection model to obtain a model score corresponding to each anomaly indicator item, perform weighted fusion of the rule score and model score corresponding to each anomaly indicator item to obtain a fusion score corresponding to each anomaly indicator item, and generate anomaly detection results for multiple accounts of the current user based on the fusion score.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device or apparatus containing the computer-readable storage medium to perform the anomaly detection method for multi-account electricity consumption data as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the communication device, the abnormal detection method for multi-account electricity consumption data as described in any one of claims 1 to 7 is implemented.

Images