一种新的约束多目标优化方法

摘要

To solve constrained multiobjective optimization problem, a double archiving mechanism was designed based on improving Strength Pareto Evolutionary Algorithm (SPEA2), which stored the excellent feasible solutions and infeasible solutions separately so as to increase the diversity of evolution and to avoid the shortcoming of designing penalty factors in a penalty function. To improve the convergence speed of the algorithm and searching ability near the constraint border, by analyzing the relationship between different constraint boundary and the Pareto front, a new method for judging the excellent infeasible solutions and environmental selection was proposed. Through simulation experiment and comparing to other algorithms,the feasibility and the advantages of this algorithm were verified.%为解决带约束的多目标优化问题,在改进的强度Pareto进化算法的基础上设计了双存档机制,分别存储进化过程中产生的优秀可行解和不可行解,增加了进化种群的多样性,避免了罚函数法需要设计惩罚系数的缺点.为提高算法的收敛速度和约束边界附近的寻优能力,通过分析不同边界与Pareto前沿的关系,提出了一种判断不可行解优劣和环境选择的新方法.通过仿真实例,并与其他算法进行比较,验证了所提算法的可行性以及在收敛速度上的优越性.