【摘要】 建立了质子交换膜燃料电池空气供给系统六阶模型。将神经网络前馈控制分别结合PID控制及线性二次型调节器（LQR）控制，来比较两种复合控制策略在燃料电池变载情况下过氧比的快速响应情况。在LQR控制策略中，首先通过单一平衡点进行模型线性化以获取状态反馈增益及降维状态观测器增益。仿真结果表明，基于单平衡点线性化的LQR结合神经网络前馈的复合控制策略在全工况范围内明显优于神经网络前馈加PID的控制策略。针对偏离平衡点较远工况处控制效果有所下降的情况，进一步采用多平衡点线性化的方法，优化设计LQR以动态调整相应增益，结果表明多平衡点LQR控制方法在过氧比控制中展现出最佳的快速响应性和稳定性。更多还原

【Abstract】 In this paper, A sixth-order model of air supply system of a proton exchange membrane fuel cell(PEMFC) was established. PID control and linear quadratic regulator(LQR) control were combined with neural network feedforward control to form two composite control strategies respectively, and the response and stability of the oxygen excess ratio(OER) during variable load were compared under two composite control strategies. In the design of the LQR controller, the model was first linearized through a single equilibrium point to obtain the state feedback gain and the reducedorder state observer gain. Simulation results show that the composite control strategy combining neural network feedforward and LQR is significantly better than the other control strategies in the full range of operating conditions, subject to linearization of the single equilibrium point.Aiming at the situation that the control effect was degraded in the working condition far away from the equilibrium point, the method of multi-equilibrium point linearization was further adopted to optimize the design of the LQR controller to dynamically adjust the corresponding gain. The results show that the multiequilibrium points LQR control method exhibits the best rapid response and stability in OER control.更多还原