Methodology for assessing system performance loss within a proactive maintenance framework

TL;DR

This paper introduces a neuro-fuzzy based methodology to evaluate system performance loss in predictive maintenance, integrating expert knowledge and real data to optimize maintenance timing.

Contribution

It proposes a novel neuro-fuzzy approach for assessing performance degradation and remaining useful life in maintenance systems, enhancing predictive accuracy.

Findings

Effective performance loss evaluation demonstrated on TELMA platform case

Neuro-fuzzy model integrates expertise and data for better predictions

Methodology supports proactive maintenance decision-making

Abstract

Maintenance plays now a critical role in manufacturing for achieving important cost savings and competitive advantage while preserving product conditions. It suggests moving from conventional maintenance practices to predictive strategy. Indeed the maintenance action has to be done at the right time based on the system performance and component Remaining Useful Life (RUL) assessed by a prognostic process. In that way, this paper proposes a methodology in order to evaluate the performance loss of the system according to the degradation of component and the deviations of system input flows. This methodology is supported by the neuro-fuzzy tool ANFIS (Adaptive Neuro-Fuzzy Inference Systems) that allows to integrate knowledge from two different sources: expertise and real data. The feasibility and added value of such methodology is then highlighted through an application case extracted from the TELMA platform used for education and research.