In this paper, we develop a mixture of Gaussians-evidential hidden Markov model (MoG-EHMM) to fuse expert knowledge and condition monitoring information for RUL prediction under the belief function theory framework. The evidential Expectation-Maximization algorithm is implemented in the offline phase to train the MoG-EHMM based on historical data. In the online phase, the trained model is used to recursively update the health state and reliability of a particular individual system. The predicted RUL is, then, represented in the form of its probability mass function. A numerical metric is defined based on the Bhattacharyya distance to measure the RUL prediction accuracy of the developed methods. We applied the developed methods on a simulation experiment and a real-world dataset from a bearing degradation test. The results demonstrate that despite imprecisions in expert knowledge, the performance of RUL prediction can be substantially improved by fusing expert knowledge with condition monitoring information.
IndexTerms-Belief function theory, expert knowledge, Mixture of Gaussians-evidential hidden Markov model (MoG-EHMM), remaining useful life (RUL) I. INTRODUCTION Prognostics and health management (PHM) has been widely recognized as a useful tool to provide early failure warnings and prevent industrial equipment from unexpected shutdowns. One of the core tasks in PHM
The fault tree analysis has been extensively implemented in failure analysis of engineered systems. In most cases, the probabilities of basic events, e.g. components’ failures, are represented by crisp values in the fault tree analyses. However, due to lack of knowledge, scarcity of failure data, or vague judgments from experts, it may produce parameter uncertainty associated with degradation models of components/systems, and such model parameter uncertainty can be quantified by the epistemic uncertainty. In addition, the common cause failure, related to the simultaneous failures of two or more components caused by physical interactions or shared environments, often exists in advanced engineered systems and computing systems. In this paper, by considering both the common cause failure and the epistemic uncertainty associated with model parameters, an evidential network model embedded with common cause failure is proposed to facilitate system failure analysis. The detailed transformations from some logic gates of a fault tree to an evidential network model are given. Moreover, the conditional belief mass tables are constructed to quantify the dependency between the states of components and the entire system. An engineering case of an aero-engine oil system, together with comparative results, is presented to demonstrate the effectiveness of the proposed evidential network model.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.