Abstract: With rapid development of China’s power utilities, power theft becomes more and more prominent, which seriously affects stable development of power supply enterprise. According to the line-loss characteristic of electricity stealing, an electricity theft identification method is presented based on similarity between user load profile and irregular line loss curve. Two methods are proposed to compute similarity of two curves in both time and frequency domain. In time domain, similarity is measured by Euclidian distance, Cosine distance and Block distance. In frequency domain, power spectrums of two curves are calculated and compared using autocorrelation method, modified covariance method and Burg method. Application results show that proposed methods can narrow electricity theft detection range and identify target accurately.

Key words: power grid, electricity theft, line loss, load profile, distance, power spectrum estimation