Electricity Theft Behavior Detection Method Based on K-means and Global Artificial Bee Colony Search Algorithm

Abstract

Electric theft behavior detection method has been a focal research topic in recent years. The traditional electricity theft detection method (ETDM) cannot detect electricity theft well because of the complicated means of stealing electricity and data analysis becomes a superior method. In this paper, a novel ETDM based on K-means and global artificial bee colony search algorithm (GABC-K-means) is proposed to address the above concerns. Artificial bee colony search algorithm (ABC) conquers local optimal resolution of K-means algorithm effectively. Global operator can adjust the dynamic balance between colony's convergence and individual's diversity and optimize global search performance. By defining the abnormal degree of load curve, we can identify the action of peccancy in using electricity. The effectiveness of the proposed method is validated in simulation section.