协变量缺失下集值映射多目标规划模型仿真  

Simulation of Multi-Objective Programming Model with Set-Valued Mapping under Missing Covariates

作　　者：吴功跃[1] 周香花[1] WU Gong-yue;ZHOU Xiang-hua(College of Science and Technology,Nanchang University,Nanchang Jiangxi 330047,China)

机构地区：[1]南昌大学科学技术学院,江西南昌330047

出　　处：《计算机仿真》2021年第11期469-472,共4页Computer Simulation

基　　金：国家自然科学基金(10461007);江西省自然科学基金(0611081)。

摘　　要：数据采集时产生的数据缺失对多目标规划结果有着直接影响,将其作为考量因素,构建一种基于协变量缺失的集值映射多目标规划模型。根据下层决策者的反应,完成上层决策者的最优决策制定,采用目标函数值集代替最优反应集,架构集值映射多目标规划问题的数学模型及对应约束条件,从模型参数与非参数两个角度,实现协变量缺失下的经验似然推断。依据不同缺失概率的经验似然检验效果、各层函数不同等级的函数评估次数以及由目标函数取得的相关指标数据等各研究成果可知,所建模型具有一定的有效性与显著的性能优越性,大幅缩减了函数评估次数,对决策变量维数有较强的鲁棒性,拥有广阔的发展前景。The lack of data during data collection has a direct impact on the results of multi-objective planning.Taking it as a consideration,a set-valued mapping multi-objective planning model based on the lack of covariates is constructed.First of all,the optimal decision of the upper decision maker was made through the response of the lower decision maker.Secondly,the optimal response set was replaced by the objective function value for constructing the mathematical model and corresponding constraints of set-valued mapping multi-objective programming.Eventually,based on the model parameters and non-parameters,the empirical likelihood inference under the absence of covariates was completed.The results show that the model established in this paper has excellent effectiveness and superiority,strong robustness and reduced function evaluation times.

关 键 词：协变量缺失 集值映射 多目标 规划模型 经验似然 随机缺失 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]