Recent advances in uncertainty analysis for prognostics and remaining useful life prediction: A review

Prognostics and health management (PHM) supports condition-based and predictive maintenance decisions of structures, systems and components. Uncertainty inevitably affects all PHM tasks of fault detection, diagnosis and prognosis. Therefore, uncertainty analysis must be performed throughout the deployment of PHM solutions for industrial practice. In this review, we systematically explore the latest research advances in uncertainty analysis for prognostics and remaining useful life (RUL) prediction, and categorize the existing approaches into three groups: statistical approaches, machine learning/deep learning approaches and hybrid approaches. For each group, the methodological advances, strengths and limitations are discussed in detail. Potential future research directions are also indicated.