Online diagnosis of PEMFC by analyzing individual cell voltages

Polymer Electrolyte Membrane Fuel Cell (PEMFC) is a promising power source for a wide range of applications. Fault diagnosis, especially online fault diagnosis, is an essential issue to promote the development and widespread use of PEMFC technology. This paper proposes a diagnosis approach for large PEMFC stack. In this approach, flooding fault is concerned; individual cell voltages are chosen as original variables for diagnosis. A dimension reduction method Fisher linear discrimination (FDA) is adopted to extract the features from the cell voltage composed vectors. After that, a classification methodology, Gaussian mixture model (GMM) is applied for fault detection. Flooding experiments were conducted on a 20-cell stack to test the approach. The obtained results showed that data points can be classified to different states of health with a high accuracy. It is also verified that the real-time implementation of the algorithm is feasible.