Abstract
This paper investigates the performance of the Non-dominated Sorting Genetic Algorithm (NSGA) based on the placement of reference points in the objective function space. An improved version of NSGA is proposed and its performance is analysed for five and eight reference points respectively in the multi-objective function space. The reference points are arranged as two effective swarm topologies: wheel and Von Neumann topology, which have been widely used in Particle Swarm Optimization (PSO). Through the simulations, the wheel topology (called wheel reference point genetic algorithm (wRPGA)) based method achieves better performance than the one which is based on the Von Neumann topology. The wheel topology also achieves better performance with respect to IGD compared to KnEA, NSGAIII and MOEAD/D for 7 out of 15 CEC 2017 benchmark problems. Moreover, wRPGA gives a good approximation of the Pareto front for the 3-objective model representing the hypothetical microgrid.