Research On Coordinated Optimization For Product Line Design And Product Platform Configuration

How to design successful product lines has become an important issue in marketing,economics,and engineering research in the last two decades.It needs to design the attribute levels of the product in a product line and,simultaneously,control the manufacturing cost,which are not addressed as two separate decisions.Therefore,many companies coordinate product line design and product platform configuration decisions within a coherent optimization framework.In order to save the cost,many companies design the product platform based on the legacy platforms,which little researches focus on.Bilevel programming is a mathematical programming whose constraints contain sub-optimization problems.It can abstract many practical issues and be widely applied.But because the upper-level contains the optimal solution or the optimal value of the lower-level,bilevel programming is a NP hard problem.Therefore,there are many experts and scholars from different fields to do some research on solutions,which can only solve some special simple bilevel programming.As a representative intelligence algorithm,genetic algorithm has a great value for solving bilevel programming problems for its better global search performance,which is a great way to solve bilevel programming.Basic on this idea,this article is to write the following several aspects:Firstly,the paper proposes a leader-follower evaluation mechanism for coordinated optimization for platform-driven product line planning.Secondly,the paper formulates an integer bilevel programming model for coordinated optimization.The upper-level addresses PLD,which decides the attribute levels of the products,while the lower-level PPC determines the configuration of each platform and the assignment of each platform to the various products in the product line.Thirdly,in order to solve this bilevel model,a bilevel genetic algorithm and an advanced encoding strategy are proposed.At last,a case study of coordinated optimization for an automobile line and product platforms is presented to demonstrate the feasibility and effectiveness of this model.by compared with the two-stage method and “all-in-one” approach.