Manufacturing nonconformities management through conditional probabilities

This paper enlightens Bayesian Networks (BNs) potentialities as a support tool, thanks to their capability of providing a graphic and intuitive representation of any process. As an engineering tool, BNs are sometimes used for reliability evaluation and in maintenance management of complex systems, but, as a matter of fact, they could be applied nearly to any field. This paper aims at illustrating how BNs can be applied to nonconformities (NCs) management. By means of a case study, we built an expert system that showed improvements both from the operations and from the strategic side. BNs were operated as an intelligent system: starting from a set of data, they were used not just as an inferential tool but the model created encoded also some human knowledge and experience, showing the added value of BN modelling. In addition to the ability of being used as an expert system, the BN was continuously improved and refined, making the model closer and closer to reality, without any amazing effort, thanks to their flexibility. As a result, we could verify the advantages of the BNs, with the addition of some information not found in the database and with the ability to quickly formalize new logical relationships of cause -effect.