抄録
In this paper, we propose a new crossover operator for real coded genetic algorithm (real-coded GA). Real-coded GA is expected to be powerful optimization technique for the nonlinear functions. A designing of crossover operator is important for a GA with good performance. At this point, for real-coded GA, many crossover operators are proposed. As a major crossover, the BLX-α proposed by Eshelman et al., the unimodal normal distribution crossover (UNDX) proposed by Ono et al.. However, the conventional crossover operators have merits and demerits of the accuracy and the processing speed. Then, in this paper, we propose the center distribution crossover (CDX) as a new crossover operator which improves the above problems. The performance of the proposed method is evaluated with the standard test functions. The performance of the proposed method are compared with the conventional methods.
