A Hybrid Probabilistic Model for Prognostic

In this paper, a hybrid probabilistic model for prognostic performance evaluation on repairable systems is proposed. This model allows to evaluate the expectation of failed components at any predetermined inspection times and over the system service life. An aeronautical application is addressed in the context of aircraft structural fatigue damage preventive maintenance. A comparative study between the hybrid model and the crude Monte Carlo method is carried out showing the efficiency of the proposed model. The expected number of failed structural components is computed at scheduled inspections and over aircraft lifetimes. The impact on the distribution of the failure threshold as well as the initial crack length one is analyzed and consequences on structural prognostic are shown. Comparative results on computing time show that the hybrid model is much less time consuming than Monte Carlo simulations. The proposed hybrid approach is flexible to use and useful to compare different maintenance policies based on both scheduled inspections or sensing SHM systems. It is therefore a relevant tool for prognosis and maintenance decision-making.