User electricity stealing behavior identification method based on non-negative matrix factorization and density clustering

Abstract

The invention relates to a user electricity stealing behavior identification method based on non-negative matrix factorization and density clustering. The method comprises the following steps: (1) preparing user electricity utilization data, including data source selection and data screening and cleaning; (2) selecting an electricity stealing behavior characteristic variable to obtain an originalelectricity stealing characteristic set; (3) extracting electricity stealing behavior characteristics based on non-negative matrix factorization; (4) establishing an improved DBSCAN electricity larceny behavior recognition model and training the model; and (5) performing electricity larceny suspicion screening on all users by utilizing the electricity larceny behavior model to obtain users with high electricity larceny suspicion degree, and checking and confirming on site by an electricity larceny inspector. Compared with a traditional electricity larceny checking mode that electricity larcenybehaviors are checked manually, the electricity larceny checking method improves the working efficiency and accuracy of electricity larceny checking, and is beneficial to reducing national electricity charge loss and reducing national property loss.

Description

Technical field [0001] The invention belongs to the technical field of anti-stealing power analysis, and specifically relates to a user's power-stealing behavior identification method based on non-negative matrix decomposition and density clustering. Background technique [0002] The theft of electricity seriously damages the legitimate rights and interests of enterprises and individuals, disrupts the normal order of power supply and use, hinders the development of the power industry, and poses a serious threat to the safe use of electricity. Tens of billions of yuan, and the indirect losses caused by accidents caused by electricity theft are even greater. [0003] On-site electricity inspection personnel currently mainly use manual methods to conduct inspections, including checking and unpacking, disassembling electric energy meters, etc., which not only requires a lot of work, but is also prone to strong opposition from users, making on-site work difficult. At present, the data ...

AI Technical Summary

Problems solved by technology

[0002] Stealing electricity seriously damages the legitimate rights and interests of enterprises and individuals, disrupts the normal order of power supply and consumption, hinders the development of electric power industry, and poses a serious threat to the safe use of electricity. In addition, the indirect losses caused by accidents caused by electricity theft are even greater

[0003] At present, on-site electricity inspectors mainly use manual methods for inspection, including checking and unpacking, dismantling electric energy meters, etc., which not only has a large workload, but also easily causes strong opposition from users, making on-site work difficult

At present, the data analysis of electricity consumption information collection is also carried out around the electric parameter data. At present, there are a large number of false alarms and false alarms in the existing abnormal data in the electricity consumption information collection system and the integrated line loss system. These noise information affect the effectiveness of the analysis. Now Some research methods are mainly based on the K-means algorithm and its variants based on the division idea, and mainly use a single algorithm. This type of algorithm cannot solve non-convex data. In the face of electricity consumption data with high information redundancy and complex electricity consumption patterns It is easy to fall into local optimum, and it is difficult to obtain ideal detection accuracy.

Images