A governance system and method based on AI collaborative rules of electricity customer profiles across the entire value chain.
By constructing a full-link governance system for AI-driven collaborative rules in electricity customer profiles, the problems of data inconsistency and insufficient automation in the governance of electricity customer profiles have been solved. This system achieves fully automated governance and improved data consistency, supporting secure data sharing between government and enterprises and its nationwide promotion.
Patent Information
- Application Number
- CN202611203113.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-10
- Publication Date
- 2026-11-17
AI Technical Summary
Existing electricity customer file management technologies cannot achieve comprehensive collaborative management, resulting in issues such as inconsistent file data, low management efficiency, and insufficient automation. Furthermore, they fail to deeply integrate the State Grid's Level 7 standard address database with authoritative government data, leading to low efficiency in processing non-standard addresses and non-standard account names.
We will build a governance system based on AI collaborative rules for electricity customer profiles, which includes a government-enterprise data security sharing module, a unified rule base and public service module, a full-dimensional profile fully automated health check module, an AI intelligent standardization module, and a full-link closed-loop rectification module. We will use a large language model for intelligent parsing and standardization to achieve fully automated closed-loop governance.
It has achieved full-dimensional automated management of electricity customer files, improved file quality and consistency, simplified the management process, and improved management efficiency and accuracy. It does not require modification of the existing marketing system and supports secure data sharing between government and enterprises and nationwide promotion.
Smart Images

Figure CN123117746A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of electricity marketing technology, and in particular to a governance system and method based on AI collaborative rules of electricity customer profiles across the entire chain. Background Technology
[0002] Basic customer files are core data assets for power grid companies in their marketing operations, providing fundamental data support for the entire business process, including business expansion applications, meter reading and verification, electricity bill collection, customer service, and emergency repair and maintenance. With the ongoing digital transformation of power marketing, the scale of customer data is constantly expanding, and various business scenarios are placing increasing demands on the completeness, accuracy, standardization, consistency, and authority of the data in these files. Therefore, customer file management has become one of the core tasks of data asset management for power grid companies.
[0003] Existing archival governance technologies mostly focus on processing single archival dimensions, failing to develop unified and synchronized governance solutions for clients' core archival elements, and thus struggling to meet the business needs of multi-dimensional collaborative archival governance. Currently, archival verification rules are deployed in different business processes across multiple heterogeneous business systems, with each system independently configuring rules and executing verifications. This results in discrepancies in verification results for the same archival document across different business processes, making it impossible to reuse governance outcomes across the entire system.
[0004] Existing technologies only perform surface-level format verification and simple cleaning for typical semi-structured archive information. They do not deeply integrate with the unified seven-level standard address database system of power grid companies, nor do they introduce authoritative government data such as housing management and market supervision for calibration. They cannot complete in-depth governance operations. A large number of non-standard addresses and non-standard household names still need to be manually verified and processed on a household-by-household basis. They cannot achieve large-scale automated governance, and there are obvious bottlenecks in governance efficiency and accuracy.
[0005] Furthermore, in the existing archive management solutions, the processes of archive verification, data cleaning, problem rectification, and result review are relatively independent. Data needs to be manually transferred between multiple systems and multiple positions, resulting in a long operation chain and a large amount of repetitive work. It has not formed a fully automated closed-loop management system from problem diagnosis, intelligent cleaning, automatic rectification to result review. Summary of the Invention
[0006] To address the shortcomings of existing technologies, this invention provides a governance system and method based on AI collaborative rules across the entire power customer profile chain. It deeply integrates the State Grid's seven-level standard address database with authoritative government data from housing management, market supervision, and other departments. It employs a large-scale artificial intelligence model and a rule engine to collaboratively drive the one-stop, full-process, automated intelligent governance of the basic profiles of power customers across all dimensions. It can be widely applied to core business scenarios such as improving the quality of customer profiles in power grid enterprises, standardizing the governance of power address usage, controlling cross-system data consistency, ensuring secure sharing of government and enterprise data, and providing data support for all marketing businesses.
[0007] On the one hand, a governance system based on AI collaborative rules of electricity customer files is provided, including: government and enterprise data security sharing module, unified rule base and public service module, full-dimensional file fully automatic physical examination module, AI intelligent standardization module, full-link closed-loop rectification module, and non-intrusive lightweight integration module; The full-dimensional archive fully automated health check module is used to receive government data and unified rule base provided by the government-enterprise data security sharing module and archive governance rules provided by the public service module, to conduct a full-dimensional diagnosis of archive quality, obtain an archive health check report, and send it to the AI intelligent standardization module and the full-link closed-loop rectification module; The AI intelligent straightening module is used to intelligently parse and straighten the file based on the problems in the physical examination report, according to the large language model finely adjusted by the power scenario, to obtain the straightened file. The end-to-end closed-loop rectification module receives the file management rules, file health check reports, and rectified files provided by the unified rule base and public service module, and reviews them. Qualified files are output to the front-end page of the existing marketing business system through a non-intrusive lightweight integration module; unqualified files are re-entered into the full-dimensional file fully automatic health check module for full-dimensional diagnosis.
[0008] Furthermore, the government-enterprise data security sharing module includes three sub-modules: a standardized government data docking sub-module, a privacy computing security interaction sub-module, and a government data dynamic update sub-module.
[0009] Furthermore, the unified rule base and public service module includes four sub-modules: rule hierarchical modeling sub-module, rule full lifecycle management sub-module, rule public service engine sub-module, and rule and government data linkage sub-module.
[0010] Furthermore, the rule-based hierarchical modeling submodule comprises a four-layer architecture: the government authority benchmark rule layer, the basic general rule layer, the scenario-specific rule layer, and the industry-specific rule layer.
[0011] Furthermore, the fully automated archive health check module comprises five sub-modules: archive data access sub-module, full-rule parallel verification sub-module, government data cross-verification sub-module, archive quality quantitative assessment sub-module, and health check report automatic generation sub-module.
[0012] Furthermore, the physical examination report automatically generated by the sub-module includes a complete list of issues, issue location, quality score, rectification priority, handling suggestions, and verification basis with authoritative government data.
[0013] Furthermore, the AI intelligent standardization module comprises four sub-modules: intelligent standardization of electricity address, intelligent standardization of account name and associated files, intelligent calibration of industry classification, and adaptive optimization of model.
[0014] Furthermore, the full-chain closed-loop rectification module includes four sub-modules: intelligent matching of rectification plans, government authority review, intelligent early warning of abnormal issues, and full-chain traceability of the governance process.
[0015] Furthermore, the non-intrusive lightweight integration module comprises three sub-modules: a browser assistant plugin sub-module, a lightweight web workbench sub-module, and a governance capability API encapsulation sub-module.
[0016] On the other hand, a governance method based on AI collaborative rules of electricity customer profiles is provided for the entire chain, including: Obtain comprehensive basic profiles of electricity customers requiring remediation; The basic files of electricity customers to be addressed are transmitted to the fully automated health check module for comprehensive diagnosis, and a health check report is obtained. Based on the problems found in the medical examination report of the archive, the AI intelligent standardization module uses a large language model finely adjusted based on the power scenario to intelligently analyze and standardize the archive based on the problems in the medical examination report, resulting in a standardized archive. The physical examination report of the archives and the rectified archives are transmitted to the full-link closed-loop rectification module for review. The qualified archives are transmitted to the front-end page of the existing marketing business system; the unqualified archives are re-entered into the full-dimensional archive fully automatic physical examination module for full-dimensional diagnosis.
[0017] The above technical solution has the following advantages or beneficial effects: This invention provides a governance system and method based on AI collaborative rules for the entire power customer profile process. It constructs a fully automated closed-loop governance architecture encompassing customer profile diagnosis, intelligent standardization, automatic rectification, authoritative government review, and end-to-end traceability. This enables end-to-end one-stop processing of customer profiles. Furthermore, it deeply integrates the State Grid's seven-level standard address database with authoritative government data from housing management and market supervision departments to build a power-specific AI intelligent standardization model. This model enables intelligent parsing and standardization of semi-structured information such as power addresses and account names. This invention requires no modification to the core logic and database of existing marketing systems, and its non-intrusive, lightweight integration architecture allows for rapid deployment and nationwide adoption of governance capabilities. Attached Figure Description
[0018] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.
[0019] Figure 1 This is a schematic diagram of the overall governance system based on AI collaborative rules of electricity customer profiles, as described in Example 1. Figure 2 This is a schematic diagram of the government-enterprise data security sharing module described in Embodiment 1; Figure 3 This is a schematic diagram of the unified rule base and public service module described in Example 1; Figure 4 This is a schematic diagram of the fully automated physical examination module for all dimensions of medical records described in Example 1; Figure 5 This is a schematic diagram of the AI intelligent straightening module described in Embodiment 1; Figure 6 This is a schematic diagram of the end-to-end closed-loop rectification module described in Example 1; Figure 7 This is a schematic diagram of the non-intrusive lightweight integrated module described in Embodiment 1; Figure 8 This is a flowchart of a governance method based on AI collaborative rules of electricity customer profiles, as described in Example 1. Detailed Implementation
[0020] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, 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 invention pertains.
[0021] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the invention. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0022] In this embodiment of the invention, "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, or B existing alone. Furthermore, in the description of this invention, "multiple" refers to two or more.
[0023] Furthermore, to facilitate a clear description of the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.
[0024] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.
[0025] All data acquisition in this embodiment is carried out in accordance with laws and regulations and with user consent, and the data is used legally.
[0026] Example 1 Currently, the management of customer basic files in domestic power grid companies generally adopts the traditional operating model of "rule-based screening, manual item-by-item verification, and batch-by-batch step-by-step rectification". This model can solve some single-dimensional file quality problems. However, with the deepening of the digital transformation of marketing business and the continuous expansion of customer data scale, the requirements for the quality, efficiency, standardization and intensification of file management are constantly increasing. The existing model can no longer meet the needs of business development.
[0027] Therefore, this embodiment provides a governance system based on AI collaborative rules of electricity customer profiles across the entire chain, such as... Figure 1 As shown, it includes: government and enterprise data security sharing module 000, unified rule base and public service module 100, full-dimensional archive fully automatic physical examination module 200, AI intelligent standardization module 300, full-link closed-loop rectification module 400, and non-intrusive lightweight integration module 500. The full-dimensional archive fully automatic health check module 200 is used to receive government data and unified rule base provided by the government-enterprise data security sharing module 000 and archive governance rules provided by the public service module 100, to perform a full-dimensional diagnosis of archive quality and obtain an archive health check report; The AI intelligent straightening module 300 is used to intelligently analyze and straighten the file based on the problems in the physical examination report, according to the large language model finely adjusted by the power scenario, to obtain the straightened file. The end-to-end closed-loop rectification module 400 is used to review the archives provided by the all-dimensional archive fully automatic physical examination module 200 and the archives regulated by the AI intelligent standardization module 300 according to the archive governance rules provided by the unified rule base and the public service module 100. Qualified archives are output to the front-end page of the existing marketing business system through the non-intrusive lightweight integration module 500; unqualified archives are re-entered to the all-dimensional archive fully automatic physical examination module 200 for all-dimensional diagnosis.
[0028] In this embodiment, the government-enterprise data security sharing module 000 serves as a basic support module, connecting to external housing management and market supervision government data sources. Its output end connects to the input end of the full-dimensional archive fully automatic physical examination module 200, and is used to provide secure and compliant government data services.
[0029] The unified rule base and public service module 100 described in this embodiment has its output end connected to the input end of the full-dimensional archive fully automatic physical examination module 200 and the full-link closed-loop rectification module 400, respectively, to provide unified rule capability support.
[0030] The fully automated archive health check module 200 described in this embodiment takes as input the customer archives to be addressed and the unqualified archives from the full-link closed-loop rectification module 400. The output is connected to the input of the AI intelligent standardization module 300 and the full-link closed-loop rectification module 400, respectively, to complete the full-dimensional diagnosis of archive quality and cross-verification of government data, and obtain the archive health check report.
[0031] The AI intelligent standardization module 300 described in this embodiment takes as input the health check report of the archives transmitted by the full-dimensional archive fully automatic health check module 200, and outputs to the input of the full-link closed-loop rectification module 400, so as to realize in-depth intelligent governance of semi-structured archives and calibration of government data, and obtain the standardized archives.
[0032] The end-to-end closed-loop rectification module 400 described in this embodiment takes as input the unified rule base and the archive management rules provided by the public service module 100, the archive examination report transmitted by the all-dimensional archive fully automatic physical examination module 200, and the archives after being regulated by the AI intelligent standardization module 300; the output is connected to the input of the non-intrusive lightweight integration module 500 to complete the closed-loop management and government authority review of the entire archive management process. Qualified archives are output to the front-end page of the existing marketing business system through the non-intrusive lightweight integration module; unqualified archives are re-entered to the all-dimensional archive fully automatic physical examination module for all-dimensional diagnosis.
[0033] The non-intrusive lightweight integration module 500 described in this embodiment communicates bidirectionally with the existing marketing business system. It transmits qualified files, as determined by the full-link closed-loop rectification module 400, to the front-end page of the existing marketing business system, and transmits customer files to be rectified in the front-end page of the existing marketing business system to the full-dimensional file fully automatic physical examination module 200.
[0034] In one embodiment, the government-enterprise data security sharing module 000 serves as the fundamental support for realizing government-enterprise collaboration, and is used to solve technical problems related to government-enterprise data connection, secure sharing, and compliant application. Figure 2As shown, it includes three sub-modules: standardized government data docking sub-module 001, privacy computing and secure interaction sub-module 002, and government data dynamic update sub-module 003.
[0035] Among them, the standardized government data docking submodule 001 is to build a standardized government data docking interface, which supports seamless docking with the government data platforms of the real estate registration center, housing management bureau, market supervision administration and industry and commerce administration, as well as the provincial government service integrated platform, to realize the timed synchronization and real-time query of authoritative data such as real estate standard address, property owner's name, enterprise registration name, unified social credit code, registered address, national standard industry classification, etc.
[0036] The privacy-preserving computing security interaction submodule 002 adopts a government-enterprise data security interaction framework built on federated learning technology. Under the premise of not disclosing original government data and not transferring original data across systems, it realizes joint verification, feature matching and model training of power archive data and authoritative government data. The data is "usable but not visible" throughout the entire process, meeting the compliance requirements of data security and personal information protection. At the same time, it builds a data desensitization, hierarchical access control and end-to-end encrypted transmission system to ensure that the data interaction is secure and controllable throughout the entire process.
[0037] The government data dynamic update submodule 003 is used to monitor authoritative data change information from the government data platform. When data such as real estate address, property owner information, enterprise registration information, and industry classification change, it automatically updates the locally cached benchmark data and triggers automatic review and rectification reminders for the corresponding power customer files to ensure the timeliness of the governance benchmark.
[0038] Furthermore, the three sub-modules of the government-enterprise data security sharing module 000 are merely logical functional divisions and do not restrict the actual deployment form: during implementation, the standardized government data docking sub-module 001 and the government data dynamic update sub-module 003 can be aggregated into an integrated government data collection and synchronization unit; alternatively, the privacy computing security interaction sub-module 002 can be extracted as an independent public security service to provide unified data security matching capabilities for other modules of the system. The module as a whole can be deployed entirely as software microservices on a cloud server, or the encryption and interface forwarding logic can be integrated into a dedicated security gateway hardware; it can adopt a centralized and unified provincial government platform, or it can complete local government-enterprise data interaction at the edge nodes of prefecture-level cities, only synchronizing a simplified index to the provincial platform. All of these methods do not deviate from the core concept of this invention and belong to equivalent implementation methods of this invention.
[0039] In one embodiment, the unified rule base and public service module 100 serves as the rule foundation for the entire system, enabling unified modeling, full lifecycle management, and cross-system public services for electricity customer profile governance rules, such as... Figure 3As shown, it includes four sub-modules: rule hierarchical modeling sub-module 101, rule full life cycle management sub-module 102, rule public service engine sub-module 103, and rule and government data linkage sub-module 104.
[0040] Among them, the rule layered modeling submodule 101 is used to build a unified rule base for the governance of customer files of power grid enterprises. It includes a four-layer architecture: government authority benchmark rule layer, basic general rule layer, scenario-specific rule layer, and industry-specific rule layer, so as to realize the refined and layered management of rules.
[0041] The government authority benchmark rule layer uses authoritative data such as standard addresses of housing management real estate, property owner information, registered names and addresses of market supervision enterprises, and national standard industry classifications as core benchmarks to construct authoritative verification rules for addresses, names, industry classifications, and certificate information, which serve as the highest priority verification benchmarks for all rules.
[0042] The basic general rules layer covers general verification rules such as mandatory fields for archives, format specifications, and coding rules stipulated by national and industry standards, including the seven-level structure specification for electricity addresses, compliance rules for household names, format rules for 11-digit mobile phone numbers, coding rules for 18-digit ID card numbers / unified social credit codes, and one-to-one correspondence rules for household registration and property change relationships.
[0043] The scenario-specific rule layer provides customized verification and governance rules for different marketing business scenarios, such as business expansion and installation, meter reading and verification, electricity bill collection, customer service, and emergency repair and maintenance.
[0044] The industry-specific rules layer is designed for different electricity consumption categories such as large industrial users, general commercial and industrial users, and residential users, and provides customized file verification and management rules.
[0045] The rule lifecycle management submodule 102 is used to realize the visual configuration, version control, dynamic update, effect evaluation and adaptive optimization of rules. It supports business personnel to add, modify, enable and disable rules through a visual interface, and record the full version change history of rules. At the same time, based on the actual application data of rules, it automatically analyzes the problem detection rate, accuracy and false judgment rate of rules, iteratively optimizes the threshold and logic of rules, and realizes the adaptive optimization of rules.
[0046] The rule public service engine submodule 103 is used to encapsulate the capabilities of the unified rule base into a standardized RESTful API service interface, providing unified rule verification, rule query, rule recommendation, and government benchmark data verification services to external heterogeneous marketing business systems and different business scenarios, realizing "one-stop maintenance, multi-scenario reuse, and multi-system calls", and completely solving the problem of inconsistent rule standards in different systems.
[0047] The rule and government data linkage submodule 104 is used to automatically synchronize and update the corresponding rule verification benchmark when the government authoritative benchmark data is updated, ensuring that the rules and government authoritative data are always consistent; at the same time, based on the verification results of government data, it automatically optimizes the priority and matching logic of the rules, continuously improving the accuracy of the rules.
[0048] This embodiment constructs a unified rule base and public service module 100 based on authoritative government data, enabling rules to be maintained in one place, managed throughout their entire lifecycle, and reused across multiple systems and scenarios, ensuring consistent verification standards across all business operations.
[0049] It should be noted that the four sub-modules under this module are only functional divisions at the logical level, not forced splits in physical deployment. There are multiple equivalent integration schemes: the rule hierarchical modeling sub-module 101 and the rule full lifecycle management sub-module 102 can be merged into a rule control and configuration unit; the rule public service engine sub-module 103 and the rule and government data linkage sub-module 104 can be integrated into a rule external service and benchmark synchronization unit; or the rule public service engine can be extracted into a global public service for the entire system to call. The entire module can be deployed independently as a dedicated rule platform, or it can be integrated as a built-in component into the power data platform and marketing platform to reuse existing base resources. The computing power carrier can be deployed centrally using general-purpose servers or distributedly using edge nodes. All of the above-mentioned splits, aggregations, and hardware / software carrier variations do not depart from the core concept of this invention and are equivalent implementations within the scope of protection of this invention's patent.
[0050] In one embodiment, the all-dimensional file fully automated health check module 200 is used to realize a one-click, all-dimensional, fully automated, parallel health check of electricity customer files, without the need for manual rule selection, and outputs a complete list of issues and quantitative quality assessment results in one go, such as Figure 4 As shown, it includes five sub-modules: archive data access sub-module 201, full-rule parallel verification sub-module 202, government data cross-verification sub-module 203, archive quality quantitative assessment sub-module 204, and physical examination report automatic generation sub-module 205.
[0051] Among them, the archive data access submodule 201 uses the non-intrusive lightweight integration module 500 to batch access the full-dimensional data of customer archives to be managed in the existing marketing business system, including core archive fields such as electricity address, account name, contact information, certificate information, industry classification, and account-transformation relationship. It also supports two modes: real-time access for single archives and offline access for batch archives.
[0052] The full-rule parallel verification submodule 202, based on the unified rule base and the unified rule base in the public service module 100, calls the distributed parallel computing engine to perform parallel verification of all rules on the archive data. It automatically completes the synchronous verification of all rules without the need for manual selection of verification rules and dimensions, which greatly improves the efficiency of physical examination.
[0053] The government data cross-verification submodule 203, through the government-enterprise data security sharing module 000, calls authoritative government data from housing management and market supervision to cross-verify the addresses, household names, enterprise names, industry classifications, and license information in the power files, accurately identifying abnormal issues that are inconsistent between the power files and authoritative government data, and marking the issue level and the basis for rectification.
[0054] The 204 sub-module for quantitative assessment of archival quality constructs a quantitative evaluation index system for archival quality across six dimensions: completeness, accuracy, standardization, consistency, authority, and timeliness. Based on the verification results, it automatically calculates the quality scores for individual archives and batches of archives. At the same time, it classifies and categorizes archival issues according to their impact on business operations and the difficulty of rectification, dividing them into three levels: immediate rectification, verification, and collaborative rectification, and matching the corresponding rectification priorities.
[0055] The automatic physical examination report generation submodule 205 automatically generates a standardized physical examination report for file quality. The physical examination report includes a complete list of problems, problem identification, quality score, rectification priority, handling suggestions, and verification basis with authoritative government data.
[0056] This embodiment constructs a fully automated health check module 200 for all-dimensional archives, enabling one-click health checks of electricity customer archives that are fully automated, parallelized, and cross-validated with government data. It eliminates the need for manual rule selection and outputs a complete list of issues and quantitative quality assessment results in one go.
[0057] It should be noted that the five sub-modules under this module are only a logical functional division, not a mandatory split in physical deployment. Multiple equivalent integration and variation implementation schemes exist. In terms of deployment, the entire module can be centrally deployed in a provincial data center computing cluster to support batch archive inspections across the province, or it can be deployed at the edge nodes of cities and counties to locally complete archive inspections within their jurisdictions. It can also be deployed independently as a specialized archive inspection service, or embedded as a functional component within marketing business systems and data governance platforms, reusing existing systems' user permissions, log auditing, and data storage infrastructure. All the above-mentioned module splitting, aggregation, and deployment carrier adjustments do not depart from the core concept of this invention and fall within the scope of equivalent implementation methods protected by patent law.
[0058] In one embodiment, the AI intelligent standardization module 300 is used to achieve deep intelligent management of semi-structured archive information such as electricity addresses and account names. It deeply integrates the State Grid's seven-level standard address database with authoritative government data, and achieves intelligent parsing and standardization based on a large language model fine-tuned specifically for the power grid company. Figure 5 As shown, it includes four sub-modules: intelligent power address organization sub-module 301, intelligent name and associated file organization sub-module 302, intelligent industry classification calibration sub-module 303, and model adaptive optimization sub-module 304.
[0059] Among them, the intelligent power address normalization submodule 301 receives the health check report of the archive transmitted by the full-dimensional archive fully automatic health check module 200. Based on the State Grid Level 7 standard address database as the basic framework and the housing management bureau standard real estate address as the authoritative calibration benchmark, it constructs a power address exclusive knowledge graph, integrates administrative division, power topology, transformer area information, and household transformer relationship data, and based on the large language model fine-tuned by the power scenario, it realizes the structured element extraction, ambiguity resolution, alias mapping, missing element completion, and format normalization of non-standard power addresses.
[0060] The specific process of the intelligent electricity address standardization submodule 301 is as follows: First, the non-standard electricity address in the health check report of the archive transmitted by the full-dimensional archive automatic health check module 200 is used as the input source. Then, based on the large language model fine-tuned by the power scenario, the non-standard address is intelligently segmented and the corresponding address elements are extracted according to the hierarchical framework of the State Grid's seven-level standard address database. Next, the extracted address elements are authoritatively matched with the standard real estate address of the Housing Management Bureau to anchor the official benchmark address. Then, based on the electricity address exclusive knowledge graph, combined with the integrated data such as administrative division, power topology, transformer area information, and household-transformer relationship, the ambiguous content in the address expression is resolved, and the mapping conversion between address aliases, colloquial names and standard names is completed. Afterwards, based on the matched standard real estate address and electricity address knowledge system, the missing hierarchical elements in the address are supplemented. Finally, a standardized electricity address with complete structure, standardized expression, and conformity to the seven-level address standard is output.
[0061] The intelligent standardization submodule 302 for household names and associated files, based on a large language model fine-tuned for the power scenario, uses the name of the property owner and the registered name of the market supervision enterprise as calibration benchmarks to achieve intelligent cleaning of household names, eliminating redundant prefixes, suffixes, special symbols and other invalid content, standardizing the writing of household names, and achieving automatic unification of abbreviations and full names; at the same time, based on the standardized address information, it links and verifies the consistency of household names, certificate information, industry classification, and household-transfer relationship, achieving linkage cleaning and standardization of multi-dimensional file information.
[0062] The Industry Classification Intelligent Calibration Submodule 303 uses the national standard industry classification issued by the Market Supervision Bureau as the authoritative benchmark. Through AI models, it automatically identifies non-standard industry classification codes and industry names within the power system, automatically matches the corresponding national standard industry classification, realizes automatic calibration of industry classification, standardization and unification of writing style, and solves the core problem of inconsistent industry classification standards in different systems.
[0063] The model adaptive optimization submodule 304 automatically performs incremental fine-tuning of the AI model based on manually reviewed results and authoritative government data, continuously improving the accuracy and recall of address resolution, name cleaning, and industry classification calibration, thereby achieving adaptive optimization of the model and obtaining the standardized archives.
[0064] This embodiment constructs an AI intelligent standardization module 300, which deeply integrates the State Grid's seven-level standard address database with authoritative government data from housing management and market supervision departments to build an AI intelligent standardization model specifically for the power scenario. This model enables intelligent parsing, element extraction, ambiguity resolution, and format normalization of semi-structured information such as electricity addresses and account names.
[0065] It should be noted that the four sub-modules under this module are only a logical functional division, not a forced split in physical deployment and code implementation. Multiple equivalent integration and variation implementation schemes exist. For example, the intelligent electricity address normalization sub-module, the intelligent account name and associated file normalization sub-module, and the intelligent industry classification calibration sub-module can be integrated into a unified intelligent file normalization inference unit, with the same model service carrying the normalization capabilities of the three types of fields. Alternatively, the model adaptive optimization sub-module can be independently extracted as a global AI model management service, providing model training, version management, and effect evaluation capabilities for other AI-related modules within the system. In terms of deployment, a centralized cloud-based large model inference architecture can be used to support batch file normalization, or, for real-time normalization scenarios at the edge, a lightweight, trimmed small model can be deployed at the city-level edge nodes, reducing computational dependence while ensuring the core normalization effect. The aforementioned algorithm replacements, sub-module aggregation and splitting, and deployment architecture adjustments, etc., do not depart from the core concept of this invention and fall within the scope of equivalent implementations protected by patent law.
[0066] In one embodiment, the end-to-end closed-loop rectification module 400 is used to achieve fully automated closed-loop management of the entire process of archive governance, constructing a complete closed loop of physical examination and diagnosis, intelligent cleaning, automatic rectification, government authority review, and end-to-end traceability, such as... Figure 6 As shown, it includes four sub-modules: intelligent matching of rectification plans (sub-module 401), government authority review (sub-module 402), intelligent early warning of abnormal issues (sub-module 403), and full-chain traceability of the governance process (sub-module 404).
[0067] Among them, the intelligent matching submodule 401 for rectification plan receives the health check report of the archives transmitted by the full-dimensional archive fully automatic health check module 200. Based on the problem classification results in the archive health check report, it automatically matches the corresponding rectification plan. For problems that need immediate rectification, such as non-standard format, inconsistent writing style, or inconsistency with government data, it triggers the automatic rectification process. For verification and collaborative problems, it pushes the manual review and collaborative handling process. At the same time, it matches the corresponding authoritative government data for each problem.
[0068] The government authority review submodule 402 automatically calls the unified rule base for a second full verification of the archives after they have been standardized by the AI intelligent standardization module 300. At the same time, it calls the government authority data through the government-enterprise data security sharing module 000 for a final verification, verifying the consistency between the rectification results and the government authority data, and outputting a list of qualified and unqualified rectification. Unqualified archives are automatically returned to the full-dimensional archive fully automatic physical examination module for the physical examination stage, forming a closed-loop management system.
[0069] The intelligent early warning submodule 403 for abnormal issues automatically triggers anomaly warnings for archival data where authoritative government data has been updated but the power system has not been updated in sync. It marks high-priority rectification tasks and pushes them to the corresponding responsible persons, enabling early detection and proactive handling of archival issues.
[0070] The 404 submodule, which tracks the entire process of archive governance, records the operation logs of the archives from access, physical examination, cleaning, rectification to review. This includes the operator, operation time, data comparison before and after rectification, rule call records, AI model call records, and government data verification records, enabling full-chain traceability of the archive governance process and meeting the requirements of power data auditing.
[0071] It should be noted that the four sub-modules under this module are only a logical functional division, not a forced split in physical deployment and code implementation. Multiple equivalent integration and variation implementation schemes exist. For example, the intelligent matching sub-module for rectification plans and the government authority review sub-module can be integrated into a unified rectification and review unit, uniformly completing rectification matching, execution, and review verification. The intelligent early warning sub-module for abnormal issues and the full-link traceability sub-module for the governance process can be merged into a control and audit unit, uniformly undertaking risk warning and compliance traceability functions. In terms of deployment, the entire module can be centrally deployed on the provincial governance platform to uniformly manage the province's archive rectification process, or it can be deployed to municipal and county-level nodes for localized handling of rectification tasks within their jurisdictions, while simultaneously storing traceability logs on the provincial platform. The entire module can also be integrated as a functional component into the power marketing work order system and data audit platform, reusing existing work order flow, access control, and audit storage infrastructure, only adding the government-linked rectification and full-link traceability logic specific to this invention. The above-mentioned sub-module aggregation and splitting, deployment architecture adjustments, and system integration changes do not deviate from the core concept of this invention and fall within the scope of equivalent implementation methods protected by patent law.
[0072] In one embodiment, the non-intrusive lightweight integration module 500 is used to achieve non-intrusive integration between the governance system and the existing marketing business system, without requiring modification to the core architecture and database of the existing marketing system, enabling rapid deployment and reuse of governance capabilities, such as... Figure 7 As shown, it contains three sub-modules: browser assistant plugin sub-module 501, lightweight web workbench sub-module 502, and governance capability API encapsulation sub-module 503.
[0073] Among them, the browser assistant plugin submodule 501 is a browser plugin developed based on Chromium and the State Grid's self-developed browser kernel. It can be directly embedded into the front-end page of the existing marketing system to realize the automatic capture and backfilling of page data. It does not require connecting to the core database of the existing marketing system. The file inspection, organization, rectification and government data verification operations can be completed on the front-end page, which meets the requirements of power data security management and control.
[0074] This plugin supports multi-core browser adaptation. Besides the Chromium-based and State Grid's self-developed browser, it is compatible with various domestically developed browsers for operating systems, meeting the requirements for information technology innovation transformation in the power industry. Functionally, it supports three operation modes: one-click system check and rectification of single files, batch capture and management of page files, and secondary confirmation of backfill results. A manual confirmation switch is set for the backfill process; changes to high-risk fields require manual confirmation by operators before being written to the front-end page, preventing business errors caused by automatic backfill.
[0075] The plugin adopts a front-end data in-memory processing mechanism throughout the process. The original archive data and governance results are not stored locally or uploaded to third-party nodes. Data interaction and backfilling are completed only in the current browser session memory, strictly adhering to the security control requirements of "minimum availability and front-end closed loop" for power data.
[0076] It also features built-in fine-grained access control and connects to the unified identity authentication system of the power grid, allowing personnel in different positions to use only the governance functions within their corresponding access scope; it supports offline caching of basic rules and address reference libraries, enabling the governance of basic format files in the event of a temporary network interruption, and synchronizing government verification results after the network is restored.
[0077] At the equivalent implementation level, in addition to browser plugins, this front-end interactive capability can also be implemented as desktop auxiliary tools, embedded components in marketing system clients, etc. As long as it adopts front-end page interaction and non-intrusive docking logic that does not intrude into the core database, it belongs to the equivalent implementation method of this module.
[0078] The lightweight Web Workbench submodule 502 is a lightweight governance workbench based on a B / S architecture. It supports batch import, governance, and export of archives, and provides functions such as visualized rule configuration, archive health check, progress monitoring, report display, and government data synchronization management. At the same time, it enables secure data interaction with existing marketing systems and data platforms through standardized API interfaces, and supports the rapid promotion and adaptation of governance capabilities.
[0079] This workbench adopts a multi-role hierarchical permission architecture, configuring different role permissions for system administrators, rule maintenance personnel, file governance operators, and audit viewers. Each role can only access corresponding functional modules and data scopes, meeting the requirements of separation of authority and responsibility for power business positions. Business functions cover the entire process of real-time governance of single files, offline governance of batch files, assignment of special governance tasks, and statistical analysis of governance effects. It supports importing and exporting file data in multiple formats such as Excel, CSV, and structured JSON. Exported files can be configured with field ranges and de-identification rules to adapt to the data flow needs of different business scenarios.
[0080] At the deployment level, it supports containerized one-click deployment, which can be quickly deployed on provincial cloud platforms, municipal local servers, or even standalone office terminals, adapting to the deployment conditions of power supply units of different sizes; it supports pure intranet offline deployment mode, with all governance capabilities running locally, and only synchronizing government baseline data and rule updates through offline data snapshots, meeting the deployment requirements of high-security intranet environments.
[0081] The workbench can be seamlessly integrated into existing marketing management system portals and data governance platform portals via single sign-on, allowing users to access the governance workbench without switching systems, further reducing system switching costs.
[0082] The governance capability API encapsulation submodule 503 encapsulates all governance capabilities, including government data verification, rule validation, and AI standardization, into standardized API interfaces, providing lightweight calling services to existing marketing business systems, 95598 customer service systems, etc., and enabling the reuse of governance capabilities across all scenarios.
[0083] This module provides multiple types of interface services, covering three categories: synchronous real-time call interfaces, asynchronous batch task interfaces, and result callback notification interfaces, which are adapted to different business call scenarios such as single real-time verification, batch file management, and event-driven triggering. The interface protocol is not limited to the RESTful architecture and can provide multiple access forms such as RPC interfaces, message queue services, and local SDK development packages to adapt to the integration needs of business systems with different technology stacks.
[0084] This embodiment designs a non-intrusive, lightweight integrated module 500 architecture, which does not require modification of the core logic and database of the existing marketing system, enabling rapid deployment and full-domain promotion of governance capabilities.
[0085] It should be noted that the three sub-modules under this module are only a logical functional division, not a mandatory separation of physical deployment and code implementation. Multiple equivalent integration and variation implementation schemes are possible. For example, the browser assistant plugin sub-module and the lightweight Web workbench sub-module can be integrated into a unified front-end interaction unit, enabling data communication and functional linkage between the plugin and the workbench. Alternatively, the governance capability API encapsulation sub-module can be extracted as an independent service gateway unit, uniformly handling the system's external capability output and security control. In terms of deployment and integration, the three sub-modules can be combined to form a complete integrated suite, or deployed separately according to business needs: the browser assistant plugin can be deployed separately for scenarios requiring only front-end page assistance; the lightweight Web workbench can be deployed separately for scenarios requiring only batch offline governance; and the encapsulated API can be called separately for scenarios requiring only embedded system governance capabilities.
[0086] The entire integrated module can serve as an independent interface layer connecting the governance system and the marketing business system, or it can be embedded as a whole into the existing marketing system and data platform, reusing existing system foundation resources as a built-in file governance capability component. The aforementioned variations in submodule aggregation and splitting, deployment and combination methods, front-end carrier replacement, and interface form adjustment do not depart from the core concept of this invention and fall within the scope of equivalent implementation methods protected by patent law.
[0087] This embodiment provides a governance system based on the full-link AI collaborative rules of electricity customer files. It addresses the specific business issues of governing the basic files of electricity customers of power grid companies by constructing an integrated technical solution with a seven-level standard address database as the core anchor, authoritative government data as the benchmark, a unified rule database as the foundation, AI and rule collaboration as dual drivers, a closed-loop governance process, non-intrusive lightweight integration as the carrier, and data security and compliance as the bottom line.
[0088] This embodiment does not replace the core data entry and verification functions of the existing marketing system. It serves only as a lightweight supplement to the existing governance model. For the first time, it deeply integrates authoritative government data such as housing management and real estate information, market supervision and enterprise information into the entire process of file governance. It completely solves the core problems of existing technologies such as "not being comprehensive, not being in-depth, not being accurate, not being able to be promoted, and not being compliant", and realizes the full-dimensional, highly authoritative, automated, standardized and intensive governance of basic files of electricity customers.
[0089] Example 2 This embodiment provides a governance method based on AI collaborative rules of electricity customer profiles across the entire chain, such as... Figure 8 As shown, it includes: Obtain comprehensive basic profiles of electricity customers requiring remediation; The basic files of electricity customers to be addressed are transmitted to the fully automated health check module for comprehensive diagnosis, and a health check report is obtained. Based on the problems found in the medical examination report of the archive, the AI intelligent standardization module uses a large language model finely adjusted based on the power scenario to intelligently analyze and standardize the archive based on the problems in the medical examination report, resulting in a standardized archive. The physical examination report of the archives and the rectified archives are transmitted to the full-link closed-loop rectification module for review. The qualified archives are transmitted to the front-end page of the existing marketing business system; the unqualified archives are re-entered into the full-dimensional archive fully automatic physical examination module for full-dimensional diagnosis.
[0090] The governance method based on AI collaborative rules of electricity customer profiles described in this embodiment includes the following specific steps: S1: Obtain comprehensive basic profiles of electricity customers to be addressed; First, standardized interfaces are used to connect with government data platforms for housing management and market supervision, and a government-enterprise data security sharing mechanism is built based on federated learning. With authoritative government data as the highest priority benchmark, a unified archive governance rule base with a four-layer architecture is built to achieve one-stop maintenance and multi-system reuse of rules.
[0091] Then, through a non-intrusive, lightweight integration module, it acquires comprehensive basic profile data of the electricity customers to be managed, supporting both single real-time access and batch offline access modes.
[0092] S2: The comprehensive basic profile of the electricity customers to be addressed is transmitted to the fully automated comprehensive profile diagnostic module for a comprehensive diagnosis, resulting in a profile diagnostic report, which specifically includes: S201: Calls the unified rule base and performs parallel automatic verification of all rules through a distributed parallel computing engine, eliminating the need for manual rule selection; S202: Access authoritative government data to cross-validate archival data and identify anomalies that are inconsistent with official authoritative data; S203: Based on a six-dimensional indicator system, calculate a quantitative score for the quality of archives, classify and categorize archive problems, and determine the priority of rectification. S204: Automatically generate archive quality inspection reports, outputting a complete list of issues, handling suggestions, and verification basis for government data.
[0093] S3: Based on the issues identified in the medical examination report, the AI intelligent restructuring module uses a large language model fine-tuned for power scenarios to intelligently analyze and restructure the file based on the problems in the medical examination report, resulting in a restructured file, specifically including: S301: For the electricity address field, based on the seven-level standard address database and the real estate address of the housing management department, address element extraction, ambiguity resolution and standardization are achieved through a large language model that has been fine-tuned for the power scenario. S302: For the household name field, the calibration benchmarks are the property owner's name and the business registration name, to achieve intelligent cleaning, redundant content removal and format standardization. S303: For the industry classification field, the national standard industry classification is used as the authoritative benchmark to achieve automatic calibration and standardization. S304: Based on standardized address information, the consistency of related files is verified in a linked manner, realizing the linkage and standardization of multi-dimensional files.
[0094] S4: Transmit the physical examination report of the file and the rectified file to the full-link closed-loop rectification module for review. The qualified files after review are transmitted to the front-end page of the existing marketing business system; the unqualified files are re-entered into the full-dimensional file fully automatic physical examination module for full-dimensional diagnosis.
[0095] S401: Based on the medical examination results and problem classification in the medical examination report, automatically match the corresponding rectification plan and government authority basis; S402: Perform automatic verification and government authoritative data review on the rectified archives to verify the rectification effect. Unqualified archives are automatically returned to the full-dimensional archive automatic physical examination module for re-examination, forming a closed loop. S403: Automatically triggers intelligent early warning and proactive rectification for archive anomalies caused by government data updates; S404: Record the entire process of archive management operation logs to achieve full-chain traceability.
[0096] The descriptions of each embodiment in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.
[0097] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A governance system based on AI collaborative rules of electricity customer profiles across the entire supply chain, characterized in that, include: The modules include: government and enterprise data security sharing module, unified rule base and public service module, full-dimensional archive fully automatic health check module, AI intelligent organization module, full-link closed-loop rectification module, and non-intrusive lightweight integration module. The full-dimensional archive fully automated health check module is used to receive government data and unified rule base provided by the government-enterprise data security sharing module and archive governance rules provided by the public service module, to conduct a full-dimensional diagnosis of archive quality, obtain an archive health check report, and send it to the AI intelligent standardization module and the full-link closed-loop rectification module; The AI intelligent straightening module is used to intelligently parse and straighten the file based on the problems in the physical examination report, according to the large language model finely adjusted by the power scenario, to obtain the straightened file. The end-to-end closed-loop rectification module receives the file management rules, file health check reports, and rectified files provided by the unified rule base and public service module, and reviews them. Qualified files are output to the front-end page of the existing marketing business system through a non-intrusive lightweight integration module; unqualified files are re-entered into the full-dimensional file fully automatic health check module for full-dimensional diagnosis.
2. The governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 1, is characterized in that... The government-enterprise data security sharing module comprises three sub-modules: a standardized government data docking sub-module, a privacy computing security interaction sub-module, and a government data dynamic update sub-module.
3. The governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 1, is characterized in that... The unified rule base and public service module comprises four sub-modules: rule hierarchical modeling sub-module, rule full lifecycle management sub-module, rule public service engine sub-module, and rule and government data linkage sub-module.
4. The governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 3, is characterized in that... The rule-based hierarchical modeling submodule comprises a four-layer architecture: government authority benchmark rule layer, basic general rule layer, scenario-specific rule layer, and industry-specific rule layer.
5. A governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 1, is characterized in that... The fully automated health check module for all-dimensional archives includes five sub-modules: archive data access sub-module, full-rule parallel verification sub-module, government data cross-verification sub-module, archive quality quantitative assessment sub-module, and health check report automatic generation sub-module.
6. A governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 5, is characterized in that... The physical examination report automatically generated by the sub-module includes a complete list of issues, issue location, quality score, rectification priority, handling suggestions, and verification basis with authoritative government data.
7. A governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 1, is characterized in that... The AI intelligent standardization module comprises four sub-modules: intelligent standardization of electricity address, intelligent standardization of account name and associated files, intelligent calibration of industry classification, and adaptive optimization of model.
8. A governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 1, is characterized in that... The end-to-end closed-loop rectification module comprises four sub-modules: intelligent matching of rectification plans, government authority review, intelligent early warning of abnormal issues, and end-to-end traceability of the governance process.
9. A governance system based on AI collaborative rules of electricity customer profiles across the entire chain, as described in claim 1, is characterized in that... The non-intrusive lightweight integration module consists of three sub-modules: a browser assistant plugin sub-module, a lightweight web workbench sub-module, and a governance capability API encapsulation sub-module.
10. A governance method based on AI collaborative rules of electricity customer profiles across the entire chain, characterized in that, include: Obtain comprehensive basic profiles of electricity customers requiring remediation; The basic files of electricity customers to be addressed are transmitted to the fully automated health check module for comprehensive diagnosis, and a health check report is obtained. Based on the problems found in the medical examination report of the archive, the AI intelligent standardization module uses a large language model finely adjusted based on the power scenario to intelligently analyze and standardize the archive based on the problems in the medical examination report, resulting in a standardized archive. The physical examination report of the archives and the rectified archives are transmitted to the full-link closed-loop rectification module for review. The qualified archives are transmitted to the front-end page of the existing marketing business system; the unqualified archives are re-entered into the full-dimensional archive fully automatic physical examination module for full-dimensional diagnosis.
