Data-driven oxygen excess ratio control for proton exchange membrane fuel cell

Efficient oxygen excess ratio (OER) control is of great importance for proton exchange membrane fuel cell because it is closely associated with the economic efficiency and safety. As widely investigated, OER control is challenging due to the difficulties of system nonlinearity, parametric uncertainty and load disturbances. In this paper, an underlying difficulty for OER control is addressed by pointing out the overshoot response. To this end, this paper employs active disturbance rejection control which is able to handle the various difficulties in a data-driven manner. It treats the nonlinearity, uncertainty and disturbances as a lumped term, which is then estimated online via analyzing the real-time data. The estimated lumped term is canceled timely such that the remaining dynamics behaves like an integrator without overshoot term therein. The data-driven and conventional proportional-integral controllers are tuned and compared based on the linearized transfer function model, showing the potential superiority of the proposed method in terms of the uncertainty and disturbance rejection, anti-windup and overshoot reduction. The nonlinear simulation based on the nonlinear mechanism model further demonstrates it good flexibility under different operating conditions. Moreover, it requires less compressor movement efforts, leading to a dynamic energy-saving effect and thus prolonging the durability and lifetime of the compressor.