Research on product variant design with uncertainty information.

During mass customization, the design specification of product and the structure of product may be uncertain because of individual customerpsilas needs. In order to implement variant design effectively in this condition, a design parameter vector model was proposed to describe the design parameters of customized product. These design parameters were distinguished as known design parameters and unknown design parameters according to the result of customerpsilas customization, and were rearranged into two different parameter subsets. Then maximum likelihood estimation method was presented to infer unknown design parameters based on known design parameters and existing mature product design specification. In order to simplify the calculation process of maximum likelihood estimation, EM algorithm was introduced to deal with this calculation process. At the same time, the main iterative steps of EM algorithm were put forward. After maximum likelihood estimation process, the value of unknown design parameters can be confirmed approximately. Therefore, the design parameter data of customized product are all known. Based on parametric CAD system, all these known design parameter data can be used to drive 3D model to variant design. So a product instance which can satisfy with customerpsilas requirement can be conducted rapidly. Finally, an example was given out to verify these principles and methods.