Application of NSGA-II Algorithm to Generation Expansion Planning

This paper describes use of a multiobjective optimization method, elitist nondominated sorting genetic algorithm version II (NSGA-II), to the generation expansion planning (GEP) problem. The proposed model provides for decision maker choice from among the different trade-off solutions. Two different problem formulations are considered. In one formulation, the first objective is to minimize cost; the second objective is to minimize sum of normalized constraint violations. In the other formulation, the first objective is to minimize investment cost; the second objective is to minimize outage cost (or maximize reliability). Virtual mapping procedure is introduced to improve the performance of NSGA-II. The GEP problem considered is a test system for a six-year planning horizon having five types of candidate units. The results are compared and validated.