PRM-based patterns for knowledge formalisation of industrial systems to support maintenance strategies assessment

Abstract

The production system and its maintenance system must be now developed on “system thinking” paradigm in order to guarantee that Key Performance Indicators (KPI) will be optimized all along the production system (operation) life. In a recursive way, maintenance system engineering has to integrate also KPI considerations with regards to its own enabling systems. Thus this paper develops a system-based methodology wherein a set of KPIs is computed in order to verify if the objectives of the production and maintenance systems are satisfied. In order to help the decision-making process for maintenance managers, a “unified” generic model have been developed. This model integrates (a) the interactions of the maintenance system with its enabling systems, (b) the impact of the maintenance strategies through the computation of some key performance indicators, and (c) different kinds of knowledge regarding the maintenance system and the system of interest, including quantitative and qualitative knowledge. This methodology is based on an executable unified model built with Probabilistic Relational Model (PRM). PRM allows a modular representation and inferences computation of large size models. The methodology added-value is shown on a test-bench.