Dual-Classifier Ensemble for Fault Detection of PEMFC

Recently, the proton exchange membrane fuel cell (PEMFC) is of increasing interest to researchers, because of its low pollution and high energy density. Fault detection is a major problem in driving the widespread use of PEMFC. Traditional classifiers are biased towards the majority classes under the condition of extremely unbalanced data distribution. This paper presents a dual-classifier ensemble algorithm for fault detection. The algorithm learns two skill-diverse classifiers by rebalancing strategy and integrates them to perform time series classification. Our comparison experiments use running data from actual PEFMC vehicles and the result of our proposed algorithm outperforms results of other algorithms.