Variation source identification methodology for multivariate nonlinear manufacturing processes

To effectively control the quality variation of manufacturing process, a variation source identification methodology for multivariate nonlinear manufacturing processes was presented. A variation source identification frame corresponded to the vector representation, part models, part variation models, and general variation source identification equation of a set of geometric feature. By using approaches of joint probability density functions, likelihood functions and likelihood ratio comparison, the judgment criterion to identify the main variation source of key characteristics of manufacturing processes was obtained. To verify the scientificity and practicality of the proposed methodology, a case study on aircraft panel components was studied by writing MATLAB program to identify the main variation source, and the result showed the feasibility of this methodology at the enterprise level.