PEM fuel cell system control: A review

Although the proton exchange membrane fuel cell (PEMFC) is still attracting enormous R&D interest because of its high energy density, its commercialization is hampered by many challenges including cutting cost, improving performance and increasing durability. While they could be solved by material selection, the durability of PEMFC is also affected by voltage reversals and fuel starvation. In this paper, PEMFC control sub-systems namely the reaction, thermal, water management and power electronic subsystems are reviewed critically, with special attention on control strategies to avoid fuel starvation. Classical proportional integral and derivative (PID) controllers are commonly used in feedback voltage control and feed-forward current control by manipulating hydrogen and air flow rates. Self-tuning PID controllers or sliding mode controllers adapt to changing dynamics and respond faster. Adaptive controllers (AC) such as load governors and extremum seeking controllers update control action continuously. Model predictive control (MPC) uses a PEMFC model to predict system behavior and update controller action. Recently, artificial intelligence such as neural network control (NNC), fuzzy logic control (FLC) and FLC-PID control have been used in PEMFC system control because they are simpler and cheaper to implement without heavy computational burden of the AC and MPC but produce better results.