Abstract The frequent variable load operation of Proton exchange membrane fuel cell (PEMFC) may cause the mismatch with the hysteresis of gas flow and diffusion, which will lead to oxygen starvation in the stack. To improve this phenomenon, a PEMFC system model was established and simplified to a linear time-invariant system, and the optimal oxygen excess ratio (OER) curve under different currents was fitted with the control objective of maximal net power in this study. Based on this, this study compared the transient control performance under PID control and model predictive control (MPC), then unprecedentedly proposed the series and parallel coupling control algorithm of MPC and PID, respectively. Simulation results manifested that the OER responds rapidly under PID control, but along with the serious overshoot and oscillation. Conversely, MPC possesses a stable control effect owing to the ability to predict and constrain the system state. More importantly, under the parallel coupling control, the PEMFC system distinguishes itself in OER tracking, output power, and output voltage by critical superiorities of stability and response speed, in which, the error integrals ISE, IAE, ITAE, and ITSE of OER tracking reach the minimum with 0.19, 0.45, 2.85, and 0.59, respectively. The output voltage fluctuations of DC-DC converter under different control modes are similar, and errors are all less than 1%, besides, the current of the DC-DC converter is more stable under the parallel coupling control. Therefore, the parallel coupling control algorithm of MPC and PID has the optimal comprehensive control effects on improving oxygen starvation.