Quality and Reliability Data Fusion for Improving Decision Making by Means of Influence Diagram: Case Study

The article focuses on the decision making in fault diagnosis of lubrication systems. In these systems, the diagnosis covers, in the majority of cases, the mechanical reliability or analysis of lubricating oils, but in a separate manner. In this section, the mechanical reliability is considered in combination with the lubricant quality, but the diagnosis process is always infected by uncertainties. Bayesian network (BN) model is developed and used as a decision-making tool. From this one, it is possible to quantify the probability of failure of this system. The diagnosis of failures is based on using Fault-Tree (FT) and Bayesian Network (BN). Firstly, a conversion from FT to BN is presented to establish a quick and accurate diagnosis. Secondly, the diagnosis is optimized by means of Influence Diagram (ID) which measures the preference.