Fault detection for PEM fuel cell using Kalman filter

Fault detection plays a key role in high cost and safe-critical systems to avoid the occurrence of an abnormal event. This work presents a model-based fault detection in the water management system of proton-exchange membrane (PEM) fuel cell stack using Kalman filter (KF). The KF is a model-based FD method relies on analytical redundancy to predict the related fault disturbed by some noise. Then, it corrected the faults with its mathematical model to minimize the error covariance between the normal and faulty scenarios by a repetitive process. Two faulty scenarios; flooding and drying which happen at anode and cathode side of PEM fuel cell are investigated. The proposed FD method using KF has successfully detected flooding and drying faults in PEMFC.