| BP神经网络在麻花钻圆度误差检测中的应用研究 |
| |
| 引用本文： | 葛动元,;姚锡凡,;向文江.BP神经网络在麻花钻圆度误差检测中的应用研究[J].武汉冶金科技大学学报,2009(4):413-417. |
| |
| 作者姓名： | 葛动元 ;姚锡凡 ;向文江 |
| |
| 作者单位： | [1]华南理工大学机械与汽车工程学院,广东广州510640; [2]邵阳学院机械与能源工程系,湖南邵阳422004 |
| |
| 基金项目： | 国家高技术研究发展计划（863）资助项目（2007AA04Z111）;湖南省教育厅重点项目（07A062）;湖南省自然科学基金资助项目（09JJ6092）. |
| |
| 摘 要： | 在麻花钻圆度误差的检测中，将BP神经网络算法引入到相应的数据处理中，以拟合出其棱边投影的椭圆表达式系数。在神经网络训练时，以钻头棱边采样点的坐标及其适当的组合作为网络的5路输入，以其输出与常数1的差值的平方为性能指标；根据梯度下降法来调整隐层神经元与输出神经元之间的连接权值，而输入层至隐层之间的连接权值不变，性能指标达到预定值时，获得一组稳定的权值，该连接权值即为钻头棱边的椭圆表达式系数；然后据此求出其较高精度的圆度误差。
|
| 关 键 词： | BP神经网络 麻花钻 圆度误差 性能指标 李雅普诺夫函数 收敛性 |
Application of BP neural network for measurement of twist-drill circularity errors |
| |
| Affiliation: | Ge Dongyuan, Yao Xifan, Xiang Wenjiang(1. School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510640, China; 2. Department of Mechanical and Energy Engineering, Shaoyang University, Shaoyang 422004.. China) |
| |
| Abstract: | BP neural network algorithm is introduced in the measurement of twist-drill circularity errors for data processing in order to fit expression coefficients of the twist-drill margin projection. The neural network is trained with coordinates of sampling points in twist-drill margin projection and their proper combinations as 5 inputs, and the square of errors between the output and constant 1 as per-formance index. The weights between hidden neurons and output neurons are tuned, while the weights between input layer and hidden layer are kept constant, and stable weight values, on behalf of expression coefficients of drill margin ellipse, are obtained in the light of gradient descent mean until desired performance index is reached. And on this basis, more precise circularity errors of the twistdrill can be solved. |
| |
| Keywords: | BP neural network twist-drill circularity convergence error performance index Lyapunov function |
| 本文献已被 维普 等数据库收录！ |
