基于BP神经网络的函数拟合算法研究
[摘要] 人工神经网络（Artificial Neural Network，ANN）是智能领域的研究
热点，目前已经成功地应用到信号处理、模式识别、机器控制、专家系统等领域中。在神经网络技术中，BP神经网络因具有结构、学习算法简单等特点，近年来得到广泛的关注，相关技术已经在预测、分类等领域中实现产业化。本文针对经典的函数拟合问题，以BP神经网络为工具，力求分析BP神经网络隐含层神经元数目对网络性能以及函数拟合效果的影响。通过Matlab的仿真实验结果表明，BP神经网络作为高效的计算智能工具，是实现复杂函数拟合的有效工具。
[关键字] 人工神经网络；BP神经网络；函数拟合
Function Fitting Based on BP Neural Network
[Abstract] Artificial neural network is a hot topic in the field of intelligent，it
has been successfully applied to the field of signal processing, pattern recognition, machine control, expert systems, etc. In the neural network technology, BP neural network due to having a structure learning algorithm and simple, it is widespread concern in recent years, related technologies have been forecasting, classification and other areas of industrialization. In this paper, a function for fitting the classic problems with BP neural network as a tool to seek to analyze the number of BP neural network hidden layer neuron impact on network performance and function fitting effect. Matlab simulation results show that, BP neural network as an efficient computational intelligence tools, is an effective tool to achieve complex function fitting
[Keywords] Artificial neural networks; BP neural network; Function fitting
目录
1 引言 ............................................................. 1
1.1 研究意义 .................................................... 1 1.2研究现状 .................................................... 1
1.2.1函数拟合的工具及方法 ................................... 2 1.2.2 神经网络的发展现状..................................... 3 1.3 本文研究内容 ................................................ 3 1.4 本文的组织结构 .............................................. 4 2 人工神经网络简介 ................................................. 5
2.1人工神经网络特性 ............................................ 5 2.2 人工神经网络的特点 .......................................... 6 2.3人工神经网络的基本特性和结构 ................................ 6 2.4 人工神经网络的主要学习算法 .................................. 8 2.5 BP神经网络结构 ............................................. 9 2.6 BP神经网络训练方法 ........................................ 11 3 基于BP神经网络的函数拟合算法设计与实现 ........................ 14
3.1面向函数拟合的BP网络结构 .................................. 14
3.1.1 函数问题的提出........................................ 14 3.1.2 建立BP神经网络....................................... 15 3.1.3 网络仿真.............................................. 15 3.1.4 网络测试.............................................. 16 3.2 实验及其结果分析 ........................................... 18
3.2.1 设置实验参数.......................................... 18 3.2.2 实验结果分析.......................................... 19
4 结论 ............................................................ 24 5 致谢语 .......................................................... 25 [参考文献] ........................................................ 26
基于BP神经网络的函数拟合算法研究
