Dynamically weighted ensemble of diverse learners for remaining useful life prediction

A traditional ensemble approach to predicting the remaining useful life (RUL) of equipment and other assets has been constructing data-driven and model-based ensembles using identical estimators. This ensemble approach may perform well on quality data collected from a design-of-experiments test but may ultimately fail when deployed in the field because of higherthan-expected noise, missing measurements, and different degradation trends. In such work environments, the high similarity of the learners can lead to large under/overestimates of RUL, where the ensemble is only as accurate as the learner which under/overestimates RUL the least. In response to this, we investigate whether an ensemble of diverse learners might be able to predict RUL consistently and accurately by aggregating the predictions of various algorithms which are found to perform differently under the same conditions. We propose improving ensemble model performance by 1) using a combination of diverse learning algorithms which are found to perform differently under the same conditions and 2) training a data-driven model to adaptively estimate the prediction weight each learner receives. The proposed methods are evaluated using open-source run-to-failure datasets from two popular systems of prognostics research: lithium-ion batteries and rolling element bearings. Results for the lithium-ion battery case study indicate that the proposed method achieves a 12% improvement in RUL prediction accuracy and a 27% decrease in the average uncertainty calibration error when compared to ensembles without dynamic weighting, and an even larger improvement when compared to individual models.