中文摘要：

针对具有不确定性的仿射非线性系统，设计了神经网络L2增益控制器，使得控制系统为有限增益L2稳定的．利用Fourier神经网络的函数拟合能力，给出了满足HJI不等式的存储函数的一般结构，并利用遗传算法对神经网络权系数进行优化，设计相应的神经网络L2增益抗干扰控制器，使得闭环系统满足相应的L2性能准则．对于如输入信号，要求控制系统设计为使得输入一输出映射为有限增益L2稳定的并有尽量小的L2增益参数．针对搅拌式化学反应器控制实例，通过数字仿真，证明此方法能够达到预期的L2性能准则．

英文摘要：

A neural network controller with L2-gain was developed for an affine nonlinear system with parameter uncertainty. The controller stabilizes the closed-loop control system with a finite L2-gain. The general structure of the storage function was formulated based on a Fourier neural network system＇s fitting capacity, which was satisfied by a Hamihonian-Jacobi inequality （HJI）. Moreover, by employing the optimization of a genetic algorithm to the weighting of the neural network system, the neural network system with its anti-disturbance system was able to meet the criteria of L2-gain performance. For an L2-gain input signal, the closed-loop control system needs to stabilize finite L2 gain to input-output mapping and have the parameters of L2 gain as small as possible. In a stirred-tank chemical reactor control example, simulation results demonstrated that the proposed method is feasible and can meet the criteria of L2-gain performance.