关键词: 大规模定制, 极大似然估计, 期望最大化算法, 变型设计

Abstract: Aiming at incomplete design parameters data problem in variant design of customized product, a variant design method of estimating unknown design parameters based on existing product data resources was presented. In this method, each product design scheme was corresponding to a set of complete design parameters, which was recorded as a vector. The unknown design parameters of customized product could be obtained through maximum likelihood estimation by using the known design parameters of customized product and existing product data resources. Expectation step, Maximization step (EM) algorithm was employed to simplify the calculation process of maximum likelihood estimation for design parameters, and an iterative process using EM algorithm to deduce unknown design parameters was presented as well. Then the framework of variant design system was designed, and variant design process based on this system was provided as well. Finally, an application example was given to verify the feasibility of this method.

Key words: mass customization, maximum likelihood estimation, expectation step, maximization step algorithm, variant design