Epistemic space of degradation processes

In this article, we present a new approach of modelling epistemic uncertainties in degradation processes. This approach is established in the framework of finite degradation structures (FDSs), which is recently proposed by the authors and can be seen as a formal extension of the fault tree analysis into multistate systems. When epistemic uncertainties are added to the states of the system, it implicitly increases the number of states and make even the Boolean systems become multistate. In the existing approaches, the addition of epistemically uncertain states as well as the new valuation mappings of the operations for those states should be done manually by the analysts depending on the type of system and its components. This manual addition may be time-consuming, error-prone and lack of generality, especially when systems get large and complex. Instead of manually remodelling the system, we propose in this article to automatic transform the model built on FDSs into epistemic space to take into account epistemic uncertainties. The proposed automatic transformations are mathematically defined and explained. As results, the uncertainty-embedded (critical) scenarios and probabilistic indicators like the belief and the plausibility in the Dempter–Shafer theory can be obtained. Illustrative examples with experimental results are also provided.