Robust prognosability for a set of partially observed discrete event systems

In this paper, we consider a robust failure prognosis problem for partially observed discrete event systems. Given a set of possible models, each of which has its own nonfailure specification, we consider a single prognoser such that, for all possible models, it predicts any failure prior to its occurrence. We call such a prognoser a robust prognoser. We introduce a notion of robust prognosability and show that it serves as a necessary and sufficient condition for the existence of a robust prognoser. We then present a method for verifying the robust prognosability condition. Moreover, we discuss online synthesis of a robust prognoser.