Detection method of coordinated fixed-rate electricity theft in high loss lines based on forward stepwise regression

• The impact of the 10 kV distribution network's contemporaneous line-loss model and electricity theft behavior is analyzed. • A mathematical model linking non-technical line losses to user electricity consumption is established. • Effective detection of fixed-ratio group electricity theft behavior is achieved. A method based on forward stepwise regression is proposed to detect group and fixed-proportion electricity theft on high-loss lines, thereby improving theft detection. Based on an analysis of the 10 kV distribution network’s concurrent line-loss model and the impact of electricity theft, the influencing factors of the line-loss rate are determined. A mathematical model relating non-technical line losses to user power consumption is established. Combined with linear regression theory, the forward stepwise regression method is adopted to construct an optimal linear regression model, which suppresses multicollinearity during the regression analysis and enables the detection of fixed-proportion group electricity theft. Experimental results show that the proposed method can effectively detect fixed-proportion group theft behavior. When the number of users involved in fixed-proportion theft changes, the EAUC and Emap(40) indices exhibit only minor fluctuations, with both metrics reaching maximum values above 94%. The method achieves effective identification of user electricity theft, and the detection results are consistent with the preset conditions.