A Differential Evolution Variant of NSGA II for Real World Multiobjective Optimization

摘要

This paper proposes the replacement of mutation and crossover operators of the NSGA II with a variant of differential evolution (DE). The resulting algorithm, termed NSGAII-DE, is tested on three test problems, and shown to be comparable to NSGA II. The algorithm is subsequently applied to two real world problems: (ⅰ.) a mass rapid transit scheduling problem and (ⅱ.) the optimization of inspection frequencies for power substations. For both the real world problems, NSGAII-DE is found to have generated better results based on comparative studies.