Agent-based prognostic function for multicomponents system with stochastic dependence

Prognostic and Health Management of multi-components systems is a challenging task, where different types of dependencies may exist. To tackle the complexity of such system, many distributed approaches are proposed. They deal with structural and functional dependencies and suppose that components are independent, which is not the case in reality. The transition of one component to more degraded states may affect other dependent components. In this paper we address the problem of prognosing components under stochastic dependence, we propose a prognostic based agent function in which a set of agents interact and coordinate to estimate the component remaining useful life with degradation interactions in a distributed manner. The simulation of an example taken from the literature is done on jade frame work, the results indicate that the precision of the component RUL is enhanced.