Modified non-dominated sorted differential evolution for multi-objective optimization
Zhao Li, Duo Li, Huang Xianchi, Zhou Bingxu · 2017
Differential Evolution (DE) algorithm is well known as a simple and efficient scheme for multi-objective global optimization over continuous spaces. In order to reduce the calculation complexity and the diversity sorting quality, the modified non-dominated sorted differential evolution (MNSDE) algorithm is proposed in this paper. The individual distribution is large-ranging and well-proportion in the evolution, by improving the crowding distance formula and the fake non-inferior solutions in the non-dominated sorting. The example based on the ZDT test function shows that the MNSDE has much better searching capacity in the whole evolution, with good diversity according to the individual distribution.