中文摘要：

针对只有输出信号可测且不可测状态与未知非线性函数相关,同时还受到外部扰动的非线性系统,提出了一种自适应神经网络动态面优化控制方法,旨在解决该类非线性系统的输出反馈跟踪问题。首先,构建了基于径向基函数神经网络的状态观测器,保证其观测误差可渐进收敛至0;其次,利用动态面技术设计了自适应输出反馈控制律;最后,利用最速下降法选取了最优控制参数,优化了控制输入和跟踪性能,减轻了控制参数整定的工作量。该方法能够保证闭环系统所有信号半全局一致有界,而且通过引入性能函数和跟踪误差转换,保证系统输出具有指定的跟踪性能,改善了控制系统的过渡性质。仿真结果表明：设计的状态观测器较准确地实现了对系统信号的估计,跟踪误差信号幅值始终不超过0.02,控制参数优化后的输入信号幅值明显减小,验证了该控制方法的有效性。

英文摘要：

Aiming at the nonlinear systems where only the output signals can be measured and their unpredictable states are associated with unknown nonlinear functions and affected by external disturbances,this paper proposes an adaptive neural network dynamic surface control method to solve the problem of output feedback tracking of nonlinear systems.First,the state observer based on RBF neural network is built to ensure the observation error can be gradually converged to zero;second,the adaptive output feedback control rule is designed by using dynamic surface technology;finally,using the steepest descent method and selecting the optimal control parameters,the control input and tracking performance are optimized,and hence the workload of control parameter setting is reduced. This method can ensure the semi-global uniform boundedness of all closed-loop signals,and guarantee system output with specified tracking performance through introducing the performance function and tracking error transformation,thus the transition quality of the control system is improved.Simulation results show that the designed state observer can realize accurate estimation of the system signals,the tracking error signal amplitude is controlled less than 0.02,and the input signal amplitude with optimized control parameters is decreased,verifying the effectiveness of the proposed control method.