基于观测器的非线性系统神经网络鲁棒控制

提出一类不依赖于模型的状态观测器，通过分析其根轨迹和极点要求配置合适的参数，该观测器本身是一个能提取高阶微分的高阶微分器．基于Lyapunov稳定性理论设计了使闭环系统渐近稳定，对模型变化和扰动具有鲁棒性的神经网络自适应控制器．该控制器不仅考虑了闭环系统的输出和设定输入误差的微分，而且考虑了误差的高阶微分，从而提高了控制品质．最后通过仿真例子验证了所提出理论的正确性．

Observer-based neural networks robust control for a class of nonlinear system

A differential state observer that does not rely on the model is presented. Its parameters are distributed through analyzing its root-locus and pole requirements. The observer itself is a high order differentiator that is able to extract the high order differential. Based on Lyapunov stability theory, an adaptive neural networks controller is designed, which makes the closed-loop system asymptotically stable and robust for changing of the model and the disturbance of the plant. The controller not only considers differential of the errors between the output and the given input but also considers high orders differentials of the errors, thus the control quality is improved. The simulation examples show the effectiveness of the mothed.