毕业设计（论文）-RBF神经滑模控制精选.doc

摘 要 简单的介绍了径向基函数Radial Basis Function），包括它的结构、性能、基本的模型、一些训练算法和学习算法以及MATLAB仿真。RBF径向基神经网络在工程中，尤其是各种智能控制中的应用十分广泛，其具有良好的非线性映射能力、自学习和泛化能力，是一种局部逼近网络,能任意精度逼近任意连续函数。 在此主要研究RBF网络逼近，神经网络的等效滑模控制以及滑模控制器的设计。利用其良好的逼近性能以切换函数作为RBF神经网络的输入,滑模控制器作为RBF神经网络的输出, 实现单入单出的神经滑模控制。 滑模控制与神经网络相结合,解决了控制系统跟踪性能和鲁棒性能之间的矛盾.系统中的滑模控制器保证了系统的快速跟踪性能，而神经网络具有很强的自学习功能,通过学习能够保证系统的稳定性,并且可对扰动和参数变化进行有效的抑制补偿,从而在不牺牲系统鲁棒性的同时达到削弱抖振。以上内容通过MATLAB仿真表明最终结果. 关键词：径向基函数，滑模控制，网络逼近，MATLAB仿真 RBF Sliding Mode Contral ABSTRACT Radial Basis Function (RBF-Radial Basis Function), including its structure, performance, the basic model, some training and learning algorithm and MATLAB simulation is introduced in this paper. Radial Basis Neural Networks is widely used in a project, particularly intelligent control, it has good nonlinear mapping ability, self-learning and generalization and it is a local network, can approximate any continuous function by Arbitrary precision. RBF approximation, the equivalent sliding mode control of neural network and the design of sliding controller are the main research content. For its good Approximation, we can make the switching function as the input of RBF network and a sliding controller as the output, then we can achieve a single-input single-output sliding mode control . Sliding mode control combined with neural network can resolve the contradiction between the tracking performance and Robust performance of the system. Sliding controller of the system satisfy the tracking performance. the self-learning function of neural network is very strong, it can satisfy the stability of the system，and inhibit and compensate for the disturbance and the change of the parameters effectively, so that it can weaken the buffeting without reduce the Robust performance. All above will be timulate by MATLAB to show the results. Keywords: radial basis function, sliding mode control, Network Approximation, MATLAB simulation RBF神经滑模控制 0 引言 复杂的系统，因其具有模型的不精确性或者不确定性，因而用很多通常的控制方法是难以有效的进行控制的。而人工神经网络（简称神经网络）由于其具有学