可靠性多目标优化设计的自适应行为博弈算法

Game algorithm for reliability multi-objective design optimization based on adaptive behavior

为解决可靠性多目标优化设计问题,并考虑各博弈方行为方式的动态性特点,提出一种自适应行为博弈算法。每个设计目标被视为一个博弈方,在计算随机设计变量对目标函数影响因子的基础上,结合模糊聚类将随机设计变量集分解为博弈方的策略集;建立了自适应行为规则,在新一轮博弈开始之前调整各博弈方的博弈行为,以及目标函数与各博弈方自身收益函数之间的映射关系;在经历多轮博弈后,根据收敛准则得到博弈均衡解。汽车制动器的案例设计结果表明,在计算效率方面,自适应行为博弈算法比竞争博弈与合作博弈算法具有更明显的优势。

To solve the problem of reliability multi-objective design optimization,an adaptive behavior-based game algorithm was proposed by considering the dynamic characteristics of players’behaviors in the game process.Each design objective was regarded as a player,and the random design variable set was decomposed into the strategy sets of players through the fuzzy clustering analysis for the influence factors of random design variables about the objective functions.The adaptive behavior rule was established.Then,the behavior and the mapping relationship between the objective function and the payoff function of each player were adjusted before the new round of game based on the rule.After multi-round of games,the equilibrium solution was obtained based on the convergence criterion.The design results of the automobile brake indicated that the proposed algorithm had the higher calculation efficiency by comparing with the competitive game and cooperative game.