Remaining useful life estimation for systems subject to multiple degradation mechanisms

Most recent studies on prognostics and remaining useful life (RUL) estimation for engineering systems are based on the underlying assumption that there is only one degradation / failure mechanism inherent in the system that is being tracked. This is often an overly simplistic assumption as systems in general are complex enough to be subject to degradation by multiple mechanisms that have different intensities in different individual units being operated. The system degrades under a cumulative effect of these multiple mechanisms and the number of competing mechanisms is many a times unknown as well. In this study, we shall present a particle filter based methodological framework that can be used to track and estimate the RUL distribution of systems with more than one mechanism of degradation / failure. The Akaike Information Criterion (AIC) is effectively used here to determine the number of inherent competing degradation mechanisms and ensure that the data is not over-fitted with many redundant degradation terms. Our methodology is illustrated here with a practical example of resistance degradation in lithium ion batteries.