Construction Technology of Abnormal Electricity Identification Feature Library

Abstract

The purpose of this paper is to introduce the method of constructing abnormal electricity identification feature library. Firstly, the typical power load characteristics are designed, including peak value, valley value, average value, power factor and other indicators, and stored in the database. Then, the characteristics of abnormal power load are analyzed, such as mutation, periodicity, duration, etc., and corresponding algorithms are developed to process and extract them. Finally, the obtained features are stored in the abnormal power load feature library. In the realization of the abnormal electricity identification feature library, the machine learning technology is used to