Statistical shape modeling in virtual assembly using PCA-technique

Paper in proceeding, 2012

The use of virtual assembly tools is one way to understand and improve the geometric product tolerance setting and the conditions for successful manufacturing. Recent developments enable consideration to be given to the deformability of parts when joined. However, in order to produce reliable results, the geometric faults of the mating surfaces must be correctly assumed. In this paper, statistical shape models built on the Principal Component Analysis-technique are proposed to be used to describe the part variation. A generalized model is presented. Furthermore, the underlying intentions and implications are discussed. The method is exemplified using the tool RD&T. In the presented case, a non-rigid sheet metal assembly is modeled and distorted to create a set of sample shapes from which a statistical model is built. In the result, the statistic representation bears a good resemblance to the nominal distorted model when the two are compared.