混合变量多目标优化设计的Pareto遗传算法实现  

Multiobjective Optimization Design with Mixed- Discrete Variables in Mechanical Engineering via Pareto Genetic Algorithm

作　　者：朱学军[1] 攀登[1] 王安麟[1] 张惠侨[1] 叶庆泰[1] 

机构地区：[1]上海交通大学机械工程学院,上海200030

出　　处：《上海交通大学学报》2000年第3期411-414,共4页Journal of Shanghai Jiaotong University

摘　　要：提出了一种用Pareto遗传算法来实施的带约束的多目标混合变量优化方法,得到Pareto最优解集,决策者从中可选出满足设计需要的解.该算法包括6个基本算子:选择、变异、交叉、离散变量圆整算子、小生境、Pareto集合过滤器.建立了用于多目标优化的适应度函数,使用模糊罚函数法将带约束的多目标优化问题转换为无约束优化问题,同时提出了处理混合变量多目标优化问题中离散变量的方法.A Pareto GA method to deal with multiobjective optimization problem was presented integrating Pareto GA and fuzzy penalty function method.By this method,a Pareto optimal set can been got,and from it the decision maker can choose a point which is most suitable for the problem.There are six operators in Pareto GA,which are selection,crossover,mutation,mixed- discrete variables rounding operator,Niche, Pareto set filter.Both continuous and discrete variables can be dealt with using this way.An example proved the efficiency and advantage of this method.

关 键 词：混合变量 PARETO最优 遗传算法 多目标优化设计 

分 类 号：TH122[机械工程—机械设计及理论] O224[理学—运筹学与控制论]