模拟动物行为的多目标可靠性优化设计博弈算法  

Game Algorithm of Multi-Objective Reliability Design Optimization Based on Simulating Animal Behavior

作　　者：冯嘉珍[1,2] 张建国 邱继伟[1,2] FENG Jia-zhen;ZHANG Jian-guo;QIU Ji-wei(School of Reliability and Systems Engineering,Beihang University,Beijing 100191,China;Science and Technology on Reliability and Environmental Engineering Laboratory,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学可靠性与系统工程学院,北京100191 [2]北京航空航天大学可靠性与环境工程技术重点实验室,北京100191

出　　处：《北京理工大学学报》2018年第5期449-453,共5页Transactions of Beijing Institute of Technology

基　　金：国家重点基础研究发展计划(2013CB733000);国家自然科学基金资助项目(51675026;71671009)

摘　　要：针对多目标可靠性优化设计的求解问题,结合动物行为对博弈方行为方式进行改进,提出模拟动物行为博弈算法.将各设计目标视为不同博弈方,通过随机向量映射技术将设计向量分解为各博弈方的策略集.根据行为匹配规则,将动物的"损人利己型"与"互惠合作型"行为赋予相应博弈方,各博弈方依据所模拟的行为,构建收益函数与目标函数的映射关系.模拟"互惠合作型"行为的博弈方组成联盟,与模拟"损人利己型"行为的博弈方进行多轮博弈,最终得到均衡解.车辆制动器设计结果表明,对比竞争与合作博弈,所提方法具有更高的计算效率.By reforming players＇ behaviors based on animal behavior,a simulating animal behavior-based game algorithm was proposed to solve the problem of multi-objective reliability design optimization.Every objective was regarded as a player,and the design vector was decomposed into players＇ strategy sets through random vector mapping technology.Selfish and mutually beneficial behaviors of animals were conferred to corresponding players based on the behavior matching rule.Then,the mapping relation between the payoff function and the objective function of every player was constructed through the simulated animal behaviors.The players that simulated mutually beneficial behaviors leagued together,and competed with other players that simulated selfish behaviors.The equilibrium solution was obtained after multi-round games.The design results of the vehicle brake show that the proposed algorithm is more efficient than competitive and cooperative games.

关 键 词：可靠性 多目标优化设计 动物行为 行为匹配 博弈 

分 类 号：TB114.3[理学—概率论与数理统计]