Multiobjective evolutionary optimisation for adaptive product family design

Abstract

Manufacturing enterprises are under competitive pressure to provide adequate product variety in order to meet diverse customer requirements while striving to reduce cost and time to market by employing product commonality and modularity. One successful approach to mass customisation (MC) is to design a family of product variants simultaneously to strike the optimum balance between commonality and differentiability. This paper formulates product family design as a multiobjective optimisation problem. A new method is proposed for assessing multi-level commonality at the product, module, component and even parameter levels. A multiobjective evolutionary algorithm (MEA) is proposed based on NSGA-II to solve this problem. This method uses a special scheme to represent and track the problem and its solutions. The effectiveness of the approach is first tested through a mathematical problem and then demonstrated with an industrial case of gantry crane family design. Computational experiments show favourable results and benefits of the proposed MEA-based product family design method.

Keywords:

• Mass customisation
• product platform
• product family design
• evolutionary algorithm
• NSGA-II
• multiobjective optimisation