基于线性物理规划的多目标差异演化算法  

Multi-objective Differential Evolution Algorithm Based on Linear Physical Programming

作　　者：叶国青 邢立宁[1] 沈永平[1] 

机构地区：[1]国防科技大学信息系统与管理学院,湖南长沙410073

出　　处：《系统工程》2011年第5期118-123,共6页Systems Engineering

基　　金：国家自然科学基金资助项目(70971131)

摘　　要：针对约束多目标优化问题求解困难的特点,提出了一种基于线性物理规划的多目标差异算法。在差异演化过程中,利用线性物理规划来处理约束和设计适应度函数,将约束多目标优化问题转化成无约束单目标优化问题求解。设计者只需设定各目标的偏好,即可得到相应的最优设计。决策过程变得更加简单和符合工程实际。仿真实验表明,该算法能够有效地求解多目标约束优化问题。A multi-objective differential evolution algorithm based on linear physical programming is proposed to solve the constrained multi-objective optimization problem.One critical mechanism embedded in this approach is applying linear physical programming to handle the constraints and to design the fitness function during differential evolution.Then the multi-objective nonlinear constrained optimization problem can be transformed into a single-objective nonlinear optimization problem.Once the designer＇s preferences are articulated,the corresponding optimal design can be obtained immediately.The decision-making process becomes easier and suites the realities of engineering design.The experimental result shows that the proposed approach can solve the multi-objective constrained optimization effectively.

关 键 词：约束多目标优化 差异演化 线性物理规划 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]