Modern large scale multi-station manufacturing systems require effective variation reduction to improve the final assembly dimensional quality. One critical measure is to diagnose the fault in the process using knowledge-based root cause identification, which can be very challenging due to the complexity of the system. The paper investigates the need of data-driven fault localization to enhance the diagnosability within the context of multiple-fault scenario(s) in multi-station assembly processes where multivariate measurements are used. The paper proposes three types of fault-signal transmission in assembly system and illustrates the nature of structured noise. Moreover, the impact of structured noise on the diagnosability is illustrated on two major fault isolation methods, namely, Principal Component Analysis and Independent Component Analysis. We then propose to use data-driven fault localization to reduce the structured noise effect and enhance the diagnosability. A simulation case study based on automotive panel assembly model is provided to illustrate the impact of structured noise and the need for data-driven localization.
A Test Restoration Method based on Genetic Algorithm for effective fault localization in multiple-fault programs
