INNOVATION OF BEHAVIOR STRUCTURE

GROUPING BASED ON GENETIC ALGORITHM

KIM Huichol^{1, 3}  GUO Weizhong¹  WANG Yuxin²  ZOU Huijun¹

(1. School of Mechanical Engineering, Shanghai Jiaotong University, Shanghai 200030；
2. School of Mechanical Engineering, Shanghai Tongji University, Shanghai 200092；
3. School of Construction Machinery, Pyongyang University of Mechanical Engineering, Pyongyang, DPR Korea )

Abstract: A GA-oriented computerized method is presented to group behaviors automatically during the innovation of behavior structures for computer-aided conceptual design of executive systems. The set of behaviors is separated into contradictory behaviors, the size of which determines the size of each section of a gene, and compatible ones, the size of which determines the number of sections of the gene. By quantifying the given evaluation criteria of behavior grouping, the fitness function is constructed. The behaviors are optimized to be grouped using the genetic algorithm with modified gene operations which leads to an optimal behavior structure. Examples are given to illustrate the presented method.

Key words: Executive system Computer-aided conceptual design Behavior structure Behavior grouping Genetic algorithm

CLC No: TH122

国家自然科学基金(50405017, 50275092)和上海市教委青年基金(01QN10)资助项目. Received 20060623, received in revised form 20061110