Self-adaptive Differential Evolution Based Multi-objective Optimization Incorporating Local Search and Indicator-Based Selection

As an efficient and effective evolutionary algorithm, Differential evolution (DE) has received ever-increasing attention over recent years. However, how to make DE suitable for multi-objective optimization is still worth further studying. Moreover, various means from different perspectives are promising to promote the performance of the algorithm. In this study, we propose a novel multi-objective evolutionary algorithm, ILSDEMO, which incorporates indicator-based selection and local search with a self-adaptive DE. In this algorithm, we also use orthogonal design to initialize the population. In addition, the k-nearest neighbor rule is employed to eliminate the most crowded solution while a new solution is ready to join the archive population. The performance of ILSDEMO is investigated on three test instances in terms of three indicators. Compared with NSGAII, IBEA, and DEMO, the results indicate that ILSDEMO can approximate the true Pareto front more accurately and evenly.