基于神经网络的自适应控制研究综述

神经网络与自适应控制相结合的研究，已成为智能控制的一个新的分支。自适应具有强鲁棒性，神经网络则具有良好的自学习功能和良好的容错能力，神经网络自适应控制由于较好地融合了两者的优点而具有强大的优势。该文综述了近年来神经网络自适应控制的研究现状，阐述了神经网络模型参考自适应控制及神经网络自校正控制两种典型的控触方案，并对神经网络自适应控制的应用作了介绍。在此基础上，对神经网络自适应控制存在的主要问题，如稳定性、鲁棒性及收敛性等问题作了积极有益的探讨。最后，展望了神经网络自适应控制未来的发展趋势，并指出了其研究方向。

Overview of Neural Network Adaptive Control

Research on a neural network combined with adaptive control has become one of new branches for intelligent control. The adaptive control has high robustness while the neural network has a self-learning function and fault-tolerance ability. As neural network adaptive control incorporates the said advantages, it has powerful superiority. This paper overviews comprehensively the existing situation of neural network adaptive control, describes two typical control schemes, neural network model reference adaptive control and neural network self-tuning control. Furthermore, the main applications of neural network adaptive control are introduced. On the basis of which this paper discusses its existing problems such as stability, robustness, convergence. In the end, this paper reviews the development trend and indicates the future research direction.