Dynamic Differential Evolution for Constrained Real-Parameter Optimization

Differential evolution (DE) has been shown to be a simple and effective evolutionary algorithm for global optimization both in benchmark test functions and many real-world applications. This paper introduces a dynamic differential evolution (D-DE) algorithm to solve constrained optimization problems. In D-DE, a novel mutation operator is firstly designed to prevent premature. Secondly, the scale factor F and the crossover probability CR are dynamic and adaptive to be beneficial for adjusting control parameters during the evolutionary process, especially, when done without any user interaction. Thirdly, D-DE uses orthogonal design method to generate initial population and reinitialize some solutions to replace some worse solutions during the search process. Finally, D-DE is validated on 6 benchmark test functions provided by the CEC 2006 special session on constrained real-parameter optimization. The experimental results obtained by D-DE are explained and discussed, and some conclusions are also drawn.