Generic and configurable diagnosis function based on production data stored in Manufacturing Execution System

The paper proposes a diagnosis approach corresponding to the specific MES level to provide information on the origins of a performance indicator degradation. Our key distribution is the proposal of a set of potential causes that may impact the successful completion of production operations, such as the operator stress, quality of material, equipment or recipe change and their characteristic parameters by exploiting MES historical database. We use Bayesian Network model to diagnose the potential failure causes and support effective human decisions on corrective actions (maintenance, human resource planning, recipe re-qualification, etc) by computing conditional probabilities for each suspected proposed causes.