基于MATLAB的改进BP神经网络的实现和比较.pdfVIP

27 1 计 算 技 术 与 自 动 化 Vol27, No1 2 0 0 8 3 Computing Technology and utomation Mar. 2 0 0 8 : 1003- 6199(2008) 01- 0028- 04 MATLAB BP 周黄斌, 周永华, 朱丽娟 ( , 530004) :BP 神经网络应用广泛, 且算法很多,研究多种神经网络的优势与不足, 以 在不同对象下选取 合适的神经网络算法运用M TL B来实现各种BP 神经网络的设计和训练并以函数逼近为例, 通过对 不同的BP 神经网络仿真与原函数图像的拟合,比较不同的BP 神经网络的性 , 验证新型 BP 网络的优势, 得出如何根据对象选取神经网络的结论 :BP 神经网络;M T L B;函数逼近 :TP391.9 : Implementation and Comparison of Improving BP Neural Network Based on MATLAB ZHOU Huang-bin,ZHOU Yong-hua,ZHU L-i juan ( College of Electrical Engineering, Guangxi University, Nanning 530004,China) Abstract: t present ,BP neural network is widely used, and has many algorithms. dvantages and disadvantages of various BP algorithms are studied so that people can choose suitable algorithms when different objective arefaced. Using M TL B imple- ments the design and training of improving BP neural network. nd as the approximation of function for example, approximating the emulation of the different BP neural Netwlrk with thefigure of primary function, it compares the performance of different BP neural network, dvantages of new BP network are validated, advice on how to select neural network base on objective. Ke words:BP neural network, ;M TL B;approximation of function M T L B , BP 1 , BP ( NN) 20 40 , , , , 2 BP 21 BP r (BP ) 1 P , r ,