Overload Risk Evaluation of DNs with High Proportion EVs Based on Adaptive Net-based Fuzzy Inference System

Abstract

Owing to the deepening of power reform and innovation of distribution networks (DNs), it is of significantly importance to make the load forecast accurately considering the new elements accessed to DNs, such as electric vehicles (EVs). Considering the impact of the charging load of large-scale EVs to DNs, this paper proposes a dynamic probabilistic method of forecasting EV charging load based on the temporal and spatial characteristics of EVs. Then, through simulating the historical charging load data of typical days, an adaptive net-based fuzzy inference system (ANFIS) is built to forecast the charging load of EVs utilizing the subtractive clustering method. Finally, on the basis of the trained ANFIS, the evaluation of the overload risk level of nodes EVs accessed to is realized. Simulation tests verify the superiority of the proposed method of forecasting the EV charging load and evaluating the overload risk level of nodes in DNs.