Evolutionary algorithm with fuzzy numbers for planning active distribution network

Modern approach to distribution systems planning requires developing tools able to interpret and analyze the stochastic nature of large power system problems. At the same time, it has to be able to optimize multiple, conflicting goals that appear in today’s open market environment. The presented evolutionary algorithm is based on a Multiple Vehicle Routing Problem adjusted for solving active electric distribution networks. An original approach to fuzzy number modeling is used for active power sources, consumption substations and distributed generation. This model, besides shorter computational time and less memory usage, is applicable to any form of fuzzy number shape required by the planner in modeling the stochastic nature of elements. An idea of defuzzification using pessimism is presented and compared to the solutions obtained without pessimism included. Application of the original universal fuzzy number modeling, using FER-fuzzy modeling, is demonstrated on multi-objective evolutionary algorithm and applied to real Croatian distribution network.