Orthogonal Predictive Differential Evolution

摘要: In traditional differential evolution (DE) algorithms, the perturbation direction of mutation is not sophisticatedly designed, which performs ineffectively or inefficiently for optimizing some complex and large-scale problems. This paper designs an orthogonal predictive scheme to solve this problem. The investigates landscape near individuals by using experimental design, then applies factor analysis predict a promising evolve. With clear sense search direction, efficiency DE improved. Moreover, step length proposed adaptively adjusted according effect prediction, helps balance exploration exploitation abilities DE. By employing such scheme, novel algorithm termed (OPDE) in paper. As OPDE can adopt different kinds classical schemes choosing base vector calculating vector, we further develop family including various variants. Experimental results demonstrate effectiveness high algorithm.