求解约束多目标区间优化的交互多属性决策NSGA-II算法  

Interactive multi-attribute decision-making NSGA-II for constrained multi-objective optimization with interval numbers

作　　者：陈志旺[1] 陈林[1] 白锌 杨七[1] 赵方亮[1] 

机构地区：[1]燕山大学工业计算机控制工程河北省重点实验室,河北秦皇岛066004

出　　处：《控制与决策》2015年第5期865-870,共6页Control and Decision

基　　金：河北省自然科学基金青年基金项目(F2014203099);燕山大学青年教师自主研究计划课题(13LGA006)

摘　　要：针对约束多目标区间优化问题,提出一种交互多属性决策NSGA-II算法.该算法将非线性问题线性化,定义p占优支配关系求出个体的序值,定义区间拥挤距离来区分具有相同序值个体的优劣,采用约束精英策略删除种群中不满足约束的个体.将选出的个体作为方案集,目标函数作为属性集,决策者对于各目标函数的偏好作为属性权重,构建一个多属性决策模型,在进化过程中融入该模型来选取符合决策者偏好的满意解.仿真实验验证了所提出方法的可行性和正确性.For constrained multi-objective optimization problems with interval numbers, a kind of interactive multi-attribute decision-making NSGA-Ⅱ is proposed. In this algorithm, nonlinear optimization problems are transformed into linear ones. P dominance relationship is defined to get the rank values of individuals and interval crowding distance is presented to distinguish the quality of the individuals with the same rank value. A constrained elite strategy is used to remove those individuals that do not satisfy the constraint from the population. A multi-attribute decision-making model（DMM） is built, which regards the selected individuals as the alternatives set, the objective function as the attributes set, and the preference of each objective function as the attribute weights. In the process of evolution, an interactive NSGA-Ⅱ incorporated with DMM is proposed to obtain satisfactory solutions which conform to the decision-makers preference. Simulation results show the feasibility and correctness of the proposed method.

关 键 词：多目标优化 不确定规划 多属性决策 区间数 进化算法 NSGA-Ⅱ 交互 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]