摘要
In order to incorporate the decision maker's preference into multiobjective optimization a preference-based multiobjective artificial bee colony algorithm PMABCA is proposed.In the proposed algorithm a novel reference point based preference expression method is addressed.The fitness assignment function is defined based on the nondominated rank and the newly defined preference distance.An archive set is introduced for saving the nondominated solutions and an improved crowding-distance operator is addressed to remove the extra solutions in the archive.The experimental results of two benchmark test functions show that a preferred set of solutions and some other non-preference solutions are achieved simultaneously.The simulation results of the proportional-integral-derivative PID parameter optimization for superheated steam temperature verify that the PMABCA is efficient in aiding to making a reasonable decision.
为了将决策者的偏好综合到多目标问题求解过程中,提出了一种偏好多目标蜂群优化算法PMABCA.在PM ABCA中,给出了一种新的偏好距离计算方法,基于非支配等级与偏好距离定义了适应度分配函数,并引入了归档集用于非支配解的存储.为了清除非支配集中多余的解,提出了改进的偏好拥挤距离算子.针对经典函数优化问题的计算结果表明,PMABCA可以在输出完整Pareto前端的基础上,确保输出大量符合偏好的最优解.将PMABCA应用于过热汽温控制系统PID参数优化问题,仿真结果表明,新算法的结果更便于决策者做出合理决策.
基金
The National Natural Science Foundation of China(No.51306082,51476027)