Development of a prognostic tool to perform reliability analysis

In maintenance field, prognostic is recognized as a key feature as the estimation of the remaining useful life of an equipment allows avoiding inopportune maintenance spending. However, it can be difficult to define and implement an adequate and efficient prognostic tool that includes the inherent uncertainty of the prognostic process. Within this frame, neuro-fuzzy systems are well suited for practical problems where it is easier to gather data (online) than to formalize the behavior of the system being studied. In this context, and according to real implementation restrictions, the paper deals with the definition of an evolutionary fuzzy prognostic system for which any assumption on its structure is necessary. The proposed approach outperform classical models and is well fitted to perform a priori reliability analysis and thereby optimize maintenance policies. An illustration of its performances is given by making a comparative study with an other neuro-fuzzy system that emerges from literature.