FL与ANN用于非线性逼近性能与机理比较研究.docVIP

毕业设计说明书(论文) 作 者： 高京 学 号：0707240415 学 院：自动化工程学院 班 级： 自动074 专 业：自动化 □测控技术与仪器 所 在 系：控制科学与工程 □仪器科学与技术 题 目： FL与ANN用于非线性系统逼近 机理与性能的比较研究 指导者： 夏琳琳 副教授 签字： 评阅者： 2011 年 6 月 吉 林  摘 要 智能控制是自动控制发展的高级阶段，代表着自动控制科学发展的最新进程。本文通过对模糊系统、BP神经网络、RBF神经网络结构及其逼近原理的研究，考查其能否逼近非线性函数，及其逼近精度指标。通过Matlab编程构造模糊系统、BP神经网络、RBF神经网络模型，设计典型的非线性函数。神经网络隐层节点个数设置为3、5、7、9，以求能够得到更为精确的结论。 BP神经网络最强大的应用之一就是函数逼近，即从训练样本出发，对未知函数进行非线性逼近。研究表明，通过BP人工神经网络所进行的非线性逼近，其精度，逼近时间，迭代次数都达到令人满意的结果，推而广之，将这种方法应用于更为复杂的实际曲面逼近有着更广阔的前景，有待我们进一步研究。RBF网络模拟人脑中局部调整，相互覆盖接受域的神经网络结构，是一种局部逼近网络。Intelligent control is the advanced stage of development of automatic control, automatic control represents the latest scientific development process. Intelligent control method using the nonlinear function approximation theory a long time and has matured, the thesis summarizes the basis of previous theoretical studies, focusing on the practical application, to approximate the simulation using Matlab. Based on the fuzzy system, BP neural network, RBF neural networks and approximation theory of the structure, examine its ability to approximate nonlinear function and its approximation accuracy indicators. Fuzzy system by Matlab programming constructs, BP neural network, RBF neural network model and design a typical non-linear function. The number of hidden layer neural network node is set to 3,5,7,9, in order to obtain more accurate conclusions. The paper gives the fuzzy logic system can approximate any arbitrary precision given function. BP neural network is one of the most powerful applications of function approximation, starting from the training samples, the unknown nonlinear function approximation. Studies show that the BP artificial neural network nonlinear approximation, its accuracy, approximation of the time, the number of iterations has reached a satisfactory outcome, by extension, will this method is applied to the actual