Differential evolution with dynamic niche radius strategy for multimodal optimization

Differential evolution algorithms are extremely effective in solving the single global optimization problems, but no DE variants perform competitively in solving multimodal optimization. In this paper, a differential evolution algorithm with dynamic niche radius strategy (DNRDE) is proposed, for multimodal optimization. In DNRDE, a new two-stage annealing schedule is designed to making the algorithm navigate from global exploration to local exploitation dynamically. Besides, in order to generate some high-quality individuals, the mutation is performed within each Euclidean neighborhood of the seed, and generating trial individuals by the perturbing procedure. Subsequently, we update the population according to the dynamic niche radius. Experiments have been performed on eight benchmark test functions. The results indicate that the proposed algorithm is better than the state-of-the-art multimodal algorithms.