基于动态结构自适应神经网络的非线性鲁棒跟踪控制

Nonlinear Robust Tracking Control Based on Dynamic Structure Adaptive Neural Network

针对非线性系统,提出一种将H∞鲁棒跟踪控制器与动态结构自适应神经网络相结合的组合控制方法。文中首先将系统线性化,设计H∞鲁棒跟踪控制器;然后针对系统中仍然存在的高阶非线性和未知不确定性,引入一种动态结构自适应神经网络,以对消非线性和不确定性的影响。这种自适应神经网络的隐层神经元随着跟踪误差的增大而在线增加,使得神经网络能以较少的神经元获得最佳的逼近效果,加快神经网络的运算速度,提高整个系统的动态性能。最后用飞行跟踪控制系统的示例证明本文方法是有效的。

A kind of combinatorial controller using H∞ robust control law and the dynamic structure neural network is presented for a class of complex nonlinear systems and the systems contain the external disturbance or some uncertainties. Firstly, the system is linearized at the working point, and H∞ robust control law is designed by considering the parameter perturbation and external disturbance. Then, the dynamic structure adaptive neural network is adopted to reduce the influence of the nonlinearity and uncertainty existed in the system. This kind of the network can approximate the optimal function by less hidden units. The units are increased with the tracking error increasing, thus the dynamic performance of the whole system is improved. The example shows that the flight tracking method is effective.