Detection of Abnormal Power Consumption State Based on VMD Decomposition and Random Matrix Theory

Abstract

Purposes Users’ abnormal power consumption behaviors need to be distinguished quickly and accurately. Methods An abnormal state detection model is proposed on the basis of smart meter data and data decomposition and random matrix theory, realizing the identification of users’ abnormal power consumption behaviors. The variational mode decomposition (VMD) algorithm is used to eliminate the noise of power data and the influence of noise data. The Random Matrix Theory (RMT) is combined with the Auto-Regressive Moving Average Model (ARMA) to improve the applicability of RMT to time series and realize the judgment of abnormal state of electricity consumption. Findings Taking the actual power consumption data of a certain area as an example, the method conveniency and efficiency for the case of large data samples and non-Gaussian distribution have been verified, which provides a new direction for the identification of abnormal power consumption behavior.

Keywords

• user behavior
• random matrix
• kernel density estimation
• abnormal power consumption
• data decomposition