A hybrid parallel genetic algorithm and its application to TSP
Dai Xiaoming①; Zou Runmin②; Feng Rui①; Zhang Hongyuan③; Shao Huihe①
①Dept. of Auto.,School of Electric and Info.,Shanghai Jiaotong Univ., Shanghai 200030 China;②College of Info. Sci. & Eng., Central South Unviersity Changsha 410083 China;③Dept. of Advanced Tech., Technology Center ZTE Corporation Shenzhen 518057 China
Abstract：This paper applies a multiple population Genetic Algorithm (GA) to solving the TSP (Traveling Salesman Problem). Different populations apply different mutation factors to achieve different search objects. The transition factor among the groups is used to solve the premature convergence problem under some circumstances. It accelerates search process in state space. The experimental results show that this algorithm has great advantage of convergence property over canonical genetic algorithm.
2002, Vol. 24 
