Comparative Study of Constrained Handling Techniques of Constrained Differential Evolution Algorithms Applied to Constrained Optimization Problems in Mechanical Engineering

Abstract

Constrained optimization problems in mechanical engineering are very difficult for the optimization algorithm. In 2013, an improved version of constrained differential evolution, named ArATM-ICDE was proposed to optimize the constrained optimization problem. An archiving-based adaptive trade-off model (ArATM) was constructed to handle the constraints; resulting in an algorithm referred to as ArATM-ICDE. This paper applies ArATM-ICDE to solve constraint optimization problems in mechanical engineering. We also combine the penalty technique for constraint handling into the ICDE, named Penalty-ICDE; which compares the abilities of the constraint handling techniques. Our experiments were conducted on ten widely used constraint engineering optimization problems. The experiment results proved the ArATM-ICDE to be more reliable than the Penalty-ICDE. Additionally, ArATM-ICDE consumed a lesser number of function calls than Penalty-ICDE. This paper further compared the effectiveness of ArATM-ICDE and Penalty-ICDE with eight state-of-the-art algorithms, which revealed that ArATM-ICDE and Penalty-ICDE produced solutions of higher quality than those produced by the comparative algorithms. The ArATM-ICDE also consumed less effort in its process.