Lubrication Oil Degradation Trajectory Prognosis with ARIMA and Bayesian Models

In order to predict lubrication oil degradation, it is important to analyze the degradation trajectory in detail. Lubrication oil degradation is influenced by numerous factors e.g. oil replenishment, oil filtering, operating conditions and system maintenance etc. that need to be considered for accurate degradation prediction. Degradation trajectory prediction provides the remaining useful life (RUL). Whereas the analysis of degradation influencing factors with their roles in prediction provides opportunity to extend or control the RUL. This paper analyzes the lubrication oil degradation trajectory under the influence of oil replenishment. We consider a data correction strategy and prognosis for lubrication oil degradation using the auto-regressive integrated moving-average (ARIMA) and Bayesian dynamic linear model (BDLM) approaches. Degradation data is generated using model-based simulations. The prediction models are then tested on the simulated degradation data set. This study exemplifies the method to find the underlying degradation model considering and identifying the degradation influencing factors.