A CGS-PGA based on Multi-Agent and Its application

Tinghong Zhao^x,Zibin Man¹,Guiquan Bi²,Zhijun Wang²

^xFaculty of Fluid Power and Control Engineering, Lanzhou University of Technology, Gansu, China, 730050;

E-mail: zhaoth2626@163.com

Abstract：The coarse grain size parallel genetic algorithm mean a lot of courses carry on the hereditary operation collaterally according to the hereditary law of every oneself, and exchange the outstanding individual among the courses, so, the speed and precision of its calculation all depend on that the communication and coordination among the courses is good or bad. This paper combine Multi-Agent theory with coarse grain size parallel genetic algorithm together, set up a coarse grain size parallel algorithm based on Multi-Agent. In this algorithm, each course is formed by one Agent, many Agent finishes heredity operate in population of itself, and exchange the outstanding individual between Agents. It has good communication and coordination between many Agent, thus make the course mutual rapidly , thus improved the speed of calculation greatly; In addition, this paper make certain improvement the concrete heredity operation of original coarse grain size parallel genetic algorithm, thus improved the precision of calculation greatly.

Keywords: Coarse grain size(CGS); Parallel genetic algorithm(PGA); Multi-Agent; Crossing; Variation;