Engineering optimization by means of an improved constrained differential evolution

To efficiently optimize the constrained engineering problems, in this paper, an improved constrained differential evolution (DE) method is proposed, where two improvements are presented. Firstly, to make the DE algorithm converge faster, a ranking-based mutation operator that is suitable to the constrained optimization problems is presented. Secondly, an improved dynamic diversity mechanism is proposed to maintain either infeasible or feasible solutions in the population. Combining the two improvements with the DE algorithm, the proposal is referred to as rank-iMDDE, for short. To evaluate the performance of rank-iMDDE, 24 benchmark functions presented in CEC’2006 are selected as the test suite. Moreover, five widely used constrained engineering benchmark problems and four constrained mechanical design problems from the literature are chosen to test the capability of rank-iMDDE for the engineering problems. Experimental results indicate that rank-iMDDE is able to improve the performance of DE in terms of the quality of the final solutions, the convergence rate, and the successful rate. Additionally, it can provide fairly-competitive results compared with other state-of-the-art evolutionary algorithms in both benchmark functions and engineering problems.