Multi-Objective Optimization of Product Configuration

Sisi Xuanyuan,Zhaoliang Jiang,Lalit Patil,Yan Li,Zhaoqian Li

DOI: https://doi.org/10.1115/detc2008-49871

2009-01-01

Abstract:In the context of globalization and mass customization, selecting the appropriate product configuration requires a simultaneous consideration of multiple criteria or objectives, which are in conflict with each other. The large solution space implies that analyzing each feasible solution is a combinatorial problem. Furthermore, no single optimal solution exists; on the contrary, there is a set of valid optimal solutions, i.e., the solution set is Pareto-optimal. We present the configuration problem from the perspective of using two types of attributes: static, i.e., the attributes that have pre-defined and constant values throughout the configuration process, and dynamic, i.e., attributes whose values vary according to decisions that are being made during the configuration process. We pose the product configuration as a multiobjective optimization problem requiring that multiple objective functions cannot be combined into a single objective function. We demonstrate the applicability of using Multi-Objective Genetic Algorithms (MOGA) to solve the problem and converge to a Pareto-optimal solution set from the large number of feasible solutions.