A self-adaptive strategy for controlling parameters in Differential Evolution

The Differential Evolution (DE) is a stochastic population-based search method for global optimization over continuous spaces. This paper presents an efficient strategy for self-adapting control parameters in Differential Evolution to solve real-parameter optimization problems. The proposed strategy introduces an adaptive mechanism at the individual level based on Cauchy distribution (CD) where the step length and crossover rate are self-adapted during the evolution process. This strategy is to utilize attractive features of CD, which has thick tails that enable it to generate considerable changes more frequently and to escape a local optima for multi-modal optimization problems. Detailed performance comparisons of a DE using the proposed strategy on wide range of fifteen standard benchmark test problems are carried out. The obtained results showed that the performance of the DE had been improved with the proposed self-adaptive strategy.