A digital twin approach for fault diagnosis in PEM fuel cell systems

A fault diagnosis approach based on digital twin (DT) is proposed for PEMFC. The established DT is in term of the dynamics model of PEM fuel cell, where the parameters therein are adaptively updated by virtue of sensor data. Fault diagnosis is performed by detecting and evaluating the residual vector. If the residual vector comes crosses the threshold of fault detection, fault isolation is performed utilizing the relative sensitivity of the fault residual. The feasibility of the proposed approach has been verified by the simulation results.