Build a Bayesian Network from FMECA in the Production of Automotive Parts: Diagnosis and Prediction

I. Ben Brahim (1) , Sid-Ali Addouche (1) , A. El Mhamedi (1) , Y. Boujelbene

1 QUARTZ - Laboratoire QUARTZ

Failure Modes, Effects and Criticality Analysis (FMECA) is one of the well-known methods of quality management that is used for continuous improvement in product or process design. This method uses linguistic expressions and has good information about cause-effect chains. However, it lacks probabilistic information. Transforming it into a Bayesian Network (BN) makes it possible to be used in maintenance for both diagnosis and prediction. The purpose of this paper is to develop a method that uses as much information as possible from FMECA, including frequency and detection to precisely make the configuration of the BN. To build a BN's structure from FMECA, we elaborate a tool to do it systematically. Moreover, we develop an algorithm to set the parameters of a BN obtained. Elicitation methods based on expert knowledge are used when data is not sufficient. A case study of FMECA in the automotive industry is introduced to verify the applicability of the proposed method in an industrial environment.

Keywords: FMECA; Bayesian Network; Decision Support System; Diagnosis; Prediction; Expert Knowledge FMECA; Expert Knowledge

Domains: Artificial Intelligence [cs.AI]; Chemical and Process Engineering; Modeling and Simulation; Machine Learning [stat.ML]

I. Ben Brahim, Sid-Ali Addouche, A. El Mhamedi, Y. Boujelbene. Build a Bayesian Network from FMECA in the Production of Automotive Parts: Diagnosis and Prediction. IFAC-PapersOnLine, 2019, 52 (13), pp.2572-2577. ⟨10.1016/j.ifacol.2019.11.594⟩. ⟨hal-03034479⟩