Methodologies for system-level remaining useful life prediction

While most prognostics approaches focus on accurate computation of the degradation rate and the Remaining Useful Life (RUL) of individual components, it is the rate at which the performance of subsystems and systems degrade that is of greater interest to the operators and maintenance personnel of these systems. Accurate and reliable predictions make it possible to plan the future operations of the system, optimize maintenance scheduling activities, and maximize system life. In system-level prognostics, we are interested in determining when the performance of a system will fall below pre-defined levels of acceptable performance. Our focus in this paper is on developing a comprehensive methodology for system-level prognostics under uncertainty that combines the use of an estimation scheme that tracks system state and degradation parameters, along with a prediction scheme that computes the RUL as a stochastic distribution over the life of the system. Two parallel methods have been developed for prediction: (1) methods based on stochastic simulation and (2) optimization methods, such as first order reliability method (FORM). We compare the computational complexity and the accuracy of the two prediction approaches using a case study of a system with several degrading components.