CN-121981524-A - Whole-process tracing and abnormal behavior identification method for power grid material internal collection transaction

CN121981524ACN 121981524 ACN121981524 ACN 121981524ACN-121981524-A

Abstract

The invention relates to the technical field of power material supply chain management and data processing, and discloses a method for identifying the whole process tracing and abnormal behavior of power grid material internal collection transaction, which collects multi-terminal data of the whole process of power grid material purchase, constructs a multi-dimensional and multi-coordinate axis composite data architecture and generates an internal collection data set; the method comprises the steps of inputting data into a decision rule judgment, establishing a clean and error standard data pool, executing progressive convergence closed loop of repair and re-judgment on repairable error data, monitoring the error data quantity in real time, utilizing a clustering model to mine error modes and update the decision rule in a reverse feeding mode to realize rule self-adaptive evolution, establishing a provider association network map, calculating the entity self and the associated side percolation conductivity based on percolation theory to generate a comprehensive evaluation value, and finally comparing the evaluation value with a preset section to generate a grading early warning signal and executing treatment. The invention effectively improves the data quality and realizes the deep penetration recognition and dynamic defense of the hidden association risk of the supply chain.

Inventors

YI YONGQIANG
ZHANG SHENGYU
HE WEN
ZHENG HONG
LIU YINGJUN
LI YINAN
YUAN ZAIXIN

Assignees

南方电网互联网服务有限公司

Dates

Publication Date: 20260505
Application Date: 20251229

Claims (10)

1. The method for tracking the whole flow of the power grid material internal collection transaction and identifying the abnormal behavior is characterized by comprising the following steps: s1, acquiring multi-terminal data of a full process of purchasing materials of a power grid, constructing a multi-dimensional and multi-coordinate axis composite data architecture, and generating an internal acquisition data set updated in real time by combining a time stamp; S2, inputting the internal data collection into a preset multidimensional basic data decision logic rule for analysis and judgment, generating clean data and error data, storing the clean data and the error data into a clean standard data pool and an error standard data pool respectively, carrying out repairability quantitative judgment on the error data, inputting the error data judged as repairable into a repairing process model for repairing, and carrying out the analysis and judgment on the repaired data again to form a data quality optimization closed loop; S3, detecting the data storage capacity of the error standard data pool in real time, when the data storage capacity reaches a preset storage capacity value, calling a preset clustering model to perform clustering analysis on error data, generating a clustering set label, identifying an error mode according to the clustering set label, generating optimization logic aiming at the error mode, and updating the multi-dimensional basic data decision logic rule based on the optimization logic; s4, extracting data from the clean standard data pool in real time, constructing a provider-associated network map, calculating the data percolation conductivity of a provider entity and the side data percolation conductivity between associated providers, and comprehensively comparing and analyzing the data percolation conductivity of the provider entity and the side data percolation conductivity to generate a comprehensive evaluation value; S5, comparing the comprehensive evaluation value with a preset evaluation section, generating a corresponding grading early warning signal, and executing corresponding grading transmission operation.
2. The method for tracking the whole flow of the power grid material internal collection transaction and identifying the abnormal behavior according to claim 1, wherein the multi-terminal data in the step S1 comprises the following steps: Internal data covering business data of enterprise resource planning, supplier relation management and supply chain management systems, external data covering supplier business, judicial and financial report data; market environment data covering bulk commodity prices and macro economic indicators; and the associated information data of the vendor bidding behavior track and the high-management associated relation data are covered.
3. The method for tracking the whole flow of the power grid material internal collection transaction and identifying the abnormal behavior according to claim 1, wherein in the step S2, repairability quantitative determination is performed on error data, and the method specifically comprises the following steps: respectively calculating a data integrity dimension score, a business rule compliance dimension score, an economical evaluation dimension score and a validity evaluation dimension score; Multiplying the data integrity dimension score, the business rule compliance dimension score, the economical evaluation dimension score and the timeliness evaluation dimension score with corresponding weight coefficients respectively, and adding the products to obtain a comprehensive repairability score; and when the comprehensive repairability score is larger than or equal to a preset repairability threshold value, marking the corresponding error data as repairable error data.
4. The method for identifying the whole process of the power grid material internal collection transaction and the abnormal behavior according to claim 3, wherein the data quality optimization closed loop in the step S2 is implemented by adopting a progressive convergence mechanism to calculate the data quality score after each cycle, and the specific logic of the calculation is as follows: The data quality score after the current cycle is equal to the sum of the data quality score after the previous cycle and an increment value, wherein the increment value is the product of a quality improvement coefficient brought by single repair processing and the residual improvement space of the data quality score after the previous cycle.
5. The method for identifying the whole process of the power grid material internal collection transaction and the abnormal behavior according to claim 1, wherein the step S3 of calling a preset clustering model to perform cluster analysis on error data comprises the following steps: aiming at the error data mixed by the numerical type and the classification type, a K-Prototypes clustering algorithm model is called; aiming at the error data with uneven density distribution, a DBSCAN density clustering algorithm model is called; And extracting error feature vectors according to the clustering result, generating the clustering set labels, and automatically recommending cleaning rules based on the clustering set labels as the optimization logic to update the multi-dimensional basic data decision logic rules.
6. The method for tracking and identifying abnormal behavior of power grid material internal collection transaction according to claim 1, wherein the step S4 is to calculate the percolation conductivity of the data of the provider entity itself specifically: Acquiring the conduction weight of each business dimension, wherein the business dimensions comprise a financial health dimension, a performance capability dimension and an operation stability dimension; Multiplying the index value normalized by each service dimension with the corresponding conduction weight, multiplying by a time attenuation factor, and accumulating the calculation results of all dimensions to obtain the percolation conductivity of the self data of the provider entity.
7. The method for identifying the whole process traceability and abnormal behavior of the internal collection transaction of the power grid supplies according to claim 1, wherein the step S4 is to calculate the percolation conductivity of the side data between the related suppliers, specifically: Determining an associated neighbor set of the target provider; calculating the edge conduction intensity coefficient between the target provider and each associated provider in the associated neighbor set; multiplying the edge conduction intensity coefficient by the corresponding associated provider entity self data percolation conductivity and the conduction attenuation coefficient of the risk on the associated edge, and accumulating the product results of all the associated paths to obtain the edge data percolation conductivity.
8. The method for tracking and identifying abnormal behavior of power grid material internal collection transaction according to claim 1, wherein after the step S4, the method further comprises a step of dynamically updating the percolation conductivity of the data of the provider entity in real time: and a sliding window mechanism is adopted, the data percolation conductivity of the provider entity at the previous moment, the internal index change quantity at the current moment and the external environment impact item are multiplied by corresponding weight coefficients respectively, and the products of the three are added to be used as the data percolation conductivity of the provider entity at the current moment.
9. The method for identifying the whole process of the internal collection transaction of the power grid material and the abnormal behavior according to claim 1, wherein in the step S4, the data percolation conductivity of the provider entity and the data percolation conductivity of the side data percolation conductivity are comprehensively compared and analyzed to generate a comprehensive evaluation value, and the method specifically comprises the following steps: respectively carrying out normalization processing on the data percolation conductivity of the provider entity and the percolation conductivity of the side data, which are obtained through real-time calculation; calculating the risk conduction intensity reflecting the overall risk scale, the risk conduction unbalance degree reflecting the structural abnormality and the risk conduction trend reflecting the dynamic evolution direction respectively based on the normalized provider entity self data percolation conductivity and the side data percolation conductivity; And multiplying the risk conduction intensity, the risk conduction unbalance degree and the risk conduction trend by corresponding normalized weight coefficients respectively, and adding the three products to obtain the comprehensive evaluation value.
10. The method for identifying the whole process of the power grid material internal collection transaction and the abnormal behavior according to claim 1, wherein the logic for generating the hierarchical early warning signal in the step S5 is as follows: Setting three incremental endpoint values of a preset evaluation section; generating a green normal signal when the comprehensive evaluation value is smaller than a minimum endpoint value; When the comprehensive evaluation value is larger than or equal to the minimum endpoint value and smaller than the middle endpoint value, generating a blue reminding signal; when the comprehensive evaluation value is larger than or equal to the middle endpoint value and smaller than the maximum endpoint value, generating a yellow reminding signal; And when the comprehensive evaluation value is greater than or equal to the maximum endpoint value, generating a red alarm signal.

Description

Whole-process tracing and abnormal behavior identification method for power grid material internal collection transaction Technical Field The invention relates to the technical field of power material supply chain management and data processing, in particular to a method for tracking the whole flow of power grid material internal collection transaction and identifying abnormal behaviors. Background With the deep promotion of smart power grid construction, the material purchasing business scale of power grid enterprises is continuously enlarged, and massive heterogeneous data covering the whole processes of purchasing planning, bidding and searching, contract performance, financial settlement and the like are generated. In order to improve transparency and compliance of the supply chain, various kinds of informationized management systems have been widely used. Through retrieval, china patent bulletin number CN109919761A discloses a blockchain platform and a transaction method for carrying out intelligent micro-grid transaction, external transaction data are received through a data acquisition module and are connected to an intelligent contract generation module, then the output end of the intelligent contract generation module is connected with the input end of a management module of an intelligent contract, the output end of the management module of the intelligent contract is connected with the input end of a verification module, the output end of the verification module is connected with the input end of a payment module, and finally the output end of the payment module is connected with the input end of a data publishing module, so that data are read from the data publishing module. The invention realizes a set of untampered, traceable and trusted database technical scheme by using the distributed collective operation method, can effectively promote the fair, public, transparent and standardized transaction of all parties in the electric power market, is beneficial to maintaining the balance of supply and demand of the market, and improves the safety and stability level of the electric power system. Trust and settlement problems in electric energy transactions are solved by using blockchain and intelligent contracts, but application logic is focused on tamper-proof recording and automatic execution of transaction results, rather than deep governance and risk mining of process data. In the specific scene of power grid material internal collection, the core pain point is the concealment of quality defects of multi-source heterogeneous data and abnormal behaviors of suppliers. The prior art comprises the patent, lacks a self-repairing closed-loop mechanism aiming at error data, is difficult to avoid the dilemma of garbage entering and exiting when facing material business data with format errors or logic deletions, and meanwhile, the technology is monitored by depending on static preset rules, is difficult to adaptively identify the abnormal modes of dynamic evolution such as bidder ring, string marks and the like, and further lacks analysis dimensions for penetrating a complex provider association network based on percolation theory to quantify risk conduction, so that the deep requirement of a material supply chain for intelligently identifying hidden risks cannot be met. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a method for tracing the whole flow of the power grid material internal collection transaction and identifying abnormal behavior, which solves the problems that the existing power grid material purchasing data management lacks a self-repairing closed-loop mechanism, the abnormal behavior identification depends on static rules and is difficult to adapt to a dynamic violation mode, and the hidden association conduction risk of a penetration identification supply chain cannot be effectively quantified. In order to achieve the above purpose, the invention provides a method for tracking the whole flow of power grid material internal collection transaction and identifying abnormal behaviors, which comprises the following steps: Collecting multi-terminal data of the whole process of purchasing power grid materials, constructing a multi-dimensional and multi-coordinate axis composite data architecture, and generating a real-time updated internal collection data set by combining a time stamp; Inputting the data set into a preset multi-dimensional basic data decision logic rule for analysis and judgment, generating clean data and error data, and storing the clean data and the error data into a clean standard data pool and an error standard data pool respectively; Performing repairability quantitative judgment on the error data, inputting the error data judged as repairable into a repair processing model for repair, and performing the analysis judgment on the repaired data again to form a data quality optimization closed loop; Monitoring the data storage capacit