A Modified Differential Evolution Algorithm for Constrained Optimization Problems

Many engineering design problems can be transformed into constrained optimization problems (COPs) which are usually very challenging for optimization algorithms. Differential Evolution (DE) has better global searching ability than traditional optimization algorithms. However, the drawbacks are also obvious: (1) The performance of DE depends largely on parameters. (2) Constraint handle technique (CHT) is necessary for COPs since DE is essentially an unconstrained optimization algorithm. This study sets up a stable optimization algorithm for COPs by combining Differential Evolution Algorithm with Strategy Adaptation (SADE) and a novel CHT. This paper proposes a simple and easy repair strategy (RP), which needn't to introduce other parameters or change the structure of algorithm. The specific operation is to repair the non-feasible solution by constructing sub-problems. To verify its superiority, the proposed Repair Strategy-Differential Evolution Algorithm with Strategy Adaptation (RP-SADE) was applied to the 2018 IEEE Congress on Evolutionary Computation (CEC2018) and one engineering application. The results show that the proposed RP-SADE has good comprehensive performance by considering computational efficiency and convergence accuracy.