Artificial intelligence-based field characteristic quantification anti-electricity larceny auxiliary checking method

Abstract

The invention discloses an artificial intelligence-based field characteristic quantification anti-electricity larceny auxiliary checking method. The method comprises the following steps: 1, establishing an electricity larceny behavior field characteristic library; 2, constructing a typical suspected electricity stealing scene sample library; 3, carrying out electricity larceny prevention key image acquisition and feature recognition by applying an OCR technology; 4, constructing an electricity larceny behavior field characteristic quantitative index library; step 5, constructing an electricity stealing behavior intelligent auxiliary judgment model based on field quantitative characteristics: taking on-site characteristic quantitative index libraries after on-site acquisition and identification of metering box door opening, electric energy meter cover opening, voltage and current reading, wiring acquisition devices and the like as the basis; and in combination with the standard judgment value of each index and the correlation degree of the electricity stealing behaviors, constructing an electricity stealing behavior intelligent auxiliary judgment model.

Description

technical field [0001] The invention relates to an artificial intelligence-based on-site feature quantification anti-stealing auxiliary investigation method used in the field of anti-stealing management. Background technique [0002] Electricity inspection is an act of inspection, supervision, guidance, and assistance for users to use electricity safely, economically and rationally in order to ensure normal electricity supply order and public safety. The traditional anti-stealing monitoring method is to compare and analyze the data period through the remote metering meter data, and find the voltage and wrong phase sequence to steal electricity. Some experienced managers can also find it through load analysis and on-site manual investigation. Stealing electricity by means of electric current. However, this method has poor immediacy, and it is difficult to effectively protect high-tech electricity theft means such as shunting in front of the watch, strong magnets, and remote ...

AI Technical Summary

Problems solved by technology

For the above-mentioned technical methods, the existing anti-stealing technology methods based on devices have the problems of low identification accuracy of electricity theft and the inability to provide accurate and reliable basis for on-site investigation of front-line employees, resulting in high work rate of on-site investigation. Research on anti-stealing data mining methods is mainly based on clustering algorithm, outlier algorithm and classification algorithm

Classification algorithms such as support vector machines and neural networks are not effective in processing large-scale sample sets. The algorithm model is greatly affected by training samples and requires corresponding labels for sample data, while most of the electricity consumption samples are unlabeled. In addition, it is not easy to select suitable training samples; the outlier algorithm and clustering algorithm are better for unlabeled samples, the algorithm is simple and fast, and the effect is better for large-scale data sets, but the scope of application of outlier algorithm samples Limited, clustering algorithms are greatly affected by sample outliers and noise points

Images