Adjusting Exploitation and Exploration Rates of Differential Evolution: A Novel Mutation Strategy

Differential evolution (DE) has attracted significant attention in recent years owing to its high performance in solving continuous problems. Up to now, a large number of variants of DE mutation strategy have been proposed and extensively studied. However, these mutation strategies rarely concentrate on adjusting the exploitation and exploration rates, which can make a great impact on DE’s performance. In this paper, a novel mutation strategy capable of visibly adjusting the exploration and exploitation rates is developed for DE algorithms. In the proposed mutation strategy, the difference vector is calculated according to the distance between selected individuals and the gravity center of these individuals. Then, the donor vector is generated by adding this difference vector to the base vector based on an asymmetrical normal distribution that determines the probability of the mutate direction to be inward for exploitation or outward for exploration. Experiments have been carried out on 13 classical benchmark functions and the CEC2013 test suite to study the proposed mutation strategy. The experimental results show that the exploitation and exploration rates have a considerable effect on both the algorithm’s convergence rate and the solution accuracy.