文献类型：期刊文献

英文题名：Formulation and application of weight-function-based physical programming

作者：Yuan, Yifeng[1];Ling, Zhihao[1,2];Gao, Chong[1];Cao, Jianfu[1]

机构：[1]E China Univ Sci & Technol, Sch Informat Sci & Engn, Shanghai 200237, Peoples R China;[2]Minist Educ, Key Lab Adv Control & Optimizat Chem Proc ECUST, Shanghai, Peoples R China

年份：2014

卷号：46

期号：12

起止页码：1628

外文期刊名：ENGINEERING OPTIMIZATION

收录：;EI(收录号：20143600036319);WOS:【SCI-EXPANDED(收录号:WOS:000342135100002)】；

基金：This work was supported by Major National Science and Technology [Special Project 2011ZX03005-004-02] and Shanghai Leading Academic Discipline [Project No. B504].

语种：英文

外文关键词：physical programming; preference; weight function; multi-objective optimization; Pareto solutions

摘要：Physical programming is effective in multi-objective optimization since it assists the designer to find the most preferred solution. Preference-function-based physical programming (PFPP) abandons the weighted-sum approach and its performance in generating Pareto solutions is susceptible to the transformation of pseudo-preferences. With the aim of integrating a weighted-sum approach into physical programming and generating well-distributed Pareto solutions, a weight-function-based physical programming (WFPP) method has been proposed. The approach forms a weight function for each normalized criterion and uses the variable weighted sum of all criteria as the aggregate objective function. Implementation for numerical and engineering design problems indicates that WFPP works as well as PFPP. The design process of generating Pareto solutions by WFPP is further presented, where the pseudo-preferences are allowed to transform in different ranges. Examples and results demonstrate that solutions generated by WFPP have better diversity performance than those of PFPP, especially when the pseudo-preferences are far from the true Pareto front.