燃料电池空气供应系统自适应神经网络滑模控制

聚合物电解质膜燃料电池(polymer electrolyte membrane fuel cell， PEMFC)空气供应系统易受参数不确定性和外部干扰的负面影响，难以实现高精度数学建模和鲁棒控制.设计了一种自适应神经网络滑模控制器，用于将PEMFC空气供应系统过氧比调节至其最优参考值，以维持最大系统输出净功率并避免氧饥饿.利用径向基函数神经网络在线逼近系统的未建模动态，而无需对外部干扰与模型参数摄动的界的先验信息.由Lyapunov理论分别推导出神经网络权值和滑模增益的自适应律，以保证闭环系统稳定性.仿真结果表明，所设计的控制器不仅改善了过氧比控制的动态行为，还有效减弱了控制输入的大幅超调和抖振.

Adaptive Neural Network Sliding Mode Control for the Fuel Cell Air Supply System

The air supply system of polymer electrolyte membrane fuel cell(PEMFC) is vulnerable to the negative impact of parameter uncertainties and external disturbances， and it is difficult to achieve high-precision mathematical modeling and robust control. An adaptive neural network sliding mode controller is designed to adjust the oxygen excess ratio of PEMFC air supply system to its optimal reference value so as to maintain the maximum system output net power and avoid oxygen starvation. The radial basis function(RBF) neural network is employed to approximate the dynamics of the unmodeled system online without prior information of the boundary of external disturbances and model parameter perturbations. To ensure the stability of the closed-loop system， the adaptive laws of neural network weights and sliding mode gains are derived by Lyapunov theory. The numerical simulation results demonstrate that the designed controller not only improves the dynamic behavior of the oxygen excess ratio control， but also effectively alleviates the large overshoot and chattering of the control input.