An improved differential evolution and novel crowding distance metric for multi-objective optimization

In this paper, an improved differential evolution based on hill-climbing techniques is proposed for multi-objective optimization. Multi-objective differential evolution optimizers are often trapped in local optima and converge slowly. A simple hill-climbing is employed to keep the diversity of population and escape from local optima. A novel crowding-distance computation procedure is proposed in order that the solutions in the neighborhood of the solutions with smallest and largest function values or locating in a lesser crowded region will have higher probability to be preserved. The proposed algorithm is tested on several classical MOP benchmark functions. The simulation results show that the proposed algorithm can obtain the solutions to be widely spread on the true Pareto optimal front.