Adaptive condition-based maintenance decision framework for deteriorating systems operating under variable environment and uncertain condition monitoring

The present article deals with the efficient use of different types of monitoring information in optimizing condition-based maintenance decision making for a deteriorating system operating under variable environment. The degradation phenomenon of a system is the fatigue crack growth that is modeled by a physics-based stochastic process. The environment process is assumed to be modeled by a time-homogenous Markov chain with finite state space. We suppose that the environmental condition is observed perfectly, while the crack depth can be assessed imperfectly through a non-destructive ultrasonic technique. As such, two kinds of indirect information are available on the system at each inspection time: environmental covariate and diagnostic covariate. Based on this set of information, two condition-based maintenance strategies adaptive to environmental conditions are developed. In the first one, the adaptation scheme is time-based, while in the second, it is condition-based. These maintenance strategies are compared one with another and to a classical non-adaptive one to point out the performances of each adaptation scheme and hence the appreciation of using different information sources in maintenance decision making.