基于动态神经网络的非线性系统建模及其控制

NONLINEAR DYNAMICS MODELING AND CONTROLLING BASED ON DYNAMIC NEURAL NETWORK

结合已知机理信息构造动态神经网络 ,进行了非线性动态系统的建模 ,给出了权值调整算法。利用获得的模型 ,设计了反馈线性化控制器。由训练好的网络在线提供反馈线性化所需要的信息。为了解决模型失配问题 ,采用内模控制结构来引入模型的误差反馈 ,以消除稳态误差。文中给出了仿真实例。

A nonlinear dynamics is modeled using a dynamic neural network in combination with known mechanistic properties of the process. The corresponding algorithm is given. The well trained neural network is used as a system model for feedback linearization in differential geometric control. An internal model control(IMC) structure is utilized to cope with the model plant mismatch, which guarantees the steady output of the plant in offset free. A simulation example shows the effectiveness of the strategy.