Multi-parental extension of the unimodal normal distribution crossover for real-coded genetic algorithms

摘要

The unimodal normal distribution crossover (UNDX) for the real-coded genetic algorithms (RCGA) proposed by Ono at al. shows excellent performance in optimization problems of multimodal and non-separable fitness functions in continuous search space. Through theoretical analysis of the UNDX, the authors have proposed design guidelines for crossover operators for the RCGA. In the present paper, a multi-parental extension of the UNDX is proposed so as to enhance its exploration ability. The formula of the extended UNDX called 'UNDX-m' is derived based on the aforesaid design guidelines. Further, the performance of the UNDX-m is evaluated by numerical experiments.