Research On Leader-follower Joint Optimization For Modular Product Family Design

Product family,as an effective means to achieve “Economies of scale” has a broad and profound impact on the industry.Under the time background of “Internet plus”,It is not only the core of mass customization concept but also an important means of industrial development under the new background.Because of its multi-body design,multi-level structure and multi-objective optimization features,modular product family design exist widely associated with master-slave structure optimization problem,which the single optimization cannot solve the multi-level and multi-agent optimization problem properly.Bi-level programming,as the only can describe the optimization method of multilevel and multi-agent model also got more extensive attention in recent years in the field of mathematical programming.But the existing research is still inadequate,this article is based on bi-level programming theory and method,starting with the modular product family design problem of master-slave hierarchical structure.This paper first discusses the main modular product family design related decisionmaking framework,and on this basis to build the optimization model of master-slave relation according to different specific problems.Then put forward combined with the characteristics of the model nested genetic algorithm as the solution method.Finally,a detailed study of three typical main modular product family design issue exists from the association.Finally,a detailed study was made on the problems of modular product family design of three kinds of typical master-slave associated problems.Respectively,with the full text of each chapter study of specific case studies will be explained.Paper work including: Firstly,hierarchical structure,decision-making frame and technical route are dicussed.Modular product family optimization has been extensively studied.This paper analyzes the decision mechanism with the characteristic of product family.It provides a theoretical basis for model construction and case studies.Sencondly,based on the intensive analysis of the hirtstvhical structure,collaborative optimization model can be constructed for the modular product family design with framework for decision-making.We propose a Genetic Algorithm method combined with characteristics of the product family for a bilevel mixed 0-1 non-linear programming model.Thirdly,based on the theoretical foundation and model,a globallocal bi-level programming model was built.Then associated optimization and nonaffiliated optimization are studied through cases.The results show that the associated optimization method is better.Fourthly,an associated optimization bi-level programming model is developed to leverage platform architecture(PA)and product family architecture(PFA).An interactive optimization structure with combination of evaluation metrics of both PA and PFA is given in line with bilevel programming.A bilevel nested genetic algorithm is put forward for efficient solution of the non-linear hierarchical joint optimization model.Fifthly,for the product family configuration with different decision makers in the competitive market,analyses of game relationships between multiple decision makers,leader-follower joint concept and method were proposed for product family configuration based on Stackelberg game.On the basis of leader-follower decision framework,a nonlinear 0-1 bi-level programming was given.The upper-and lower-level were separately corresponding to the leading manufacturer and following manufacturer.