Modified Differential Evolution for Dynamic Optimization Problems

Abstract

Modified differential evolution algorithm (MDE) is proposed for dynamic optimization problems. The new algorithm divides the population into two, a main subpopulation and an assistant one. The main subpopulation keeps invariant and searches locally. The assistant subpopulatioin is re-initialized at random and searches globally. The results show that MDE can track the changing extreme promptly and accurately and is capable of efficiently solving dynamic optmization problems.