Computational method for diagnosis of variation-related assembly problems

Abstract

Dimensional variations often cause expensive problems in the assembly of mechanical products especially during new product deployment. Currently, the time and labour spent on the removal of variation-related assembly problems are considerable. This paper proposes a computational method to identify the causes of variation-related problems in assemblies. In the proposed method, the relationship between causes and symptoms is computed via simulation of the assembly process, and they are represented by a matrix. For given measurement data, this method uses linear optimization and statistical techniques to diagnose assembly problems. The proposed method can diagnose assembly problems caused by either a single or multiple fault(s).