Diagnosis of Complex System in Mechanical with Multiple Failure Sources

Abstract: This paper proposes a method to accurately locate the source of product quality drift in a flexible manufacturing system (FMS). Based on the relationship between fault sources, a fault propagation mechanism in the production process is proposed to explain fault propagation after production. Logic diagnostic models are used to reduce the search space for suspicious equipment in the production process; however, this does not help to accurately locate the faulty equipment. In the proposed method, we model this reduced search space as a Bayesian network, which uses historical data to calculate the conditional probability of each suspicious device. This method helps to make accurate decisions to locate the cause of product quality drift, whether it is one of the devices in the production process or the product itself. Keywords: Fault Diagnosis, Complex Manufacturing System, Bayesian Network