北京航空航天大学学报 ›› 2011, Vol. 37 ›› Issue (4): 405-408.

• 论文 • 上一篇    下一篇

基于人工神经网络的柔性机翼挠度预测

王志飞, 王华, 贾青萍, 韩晶   

  1. 北京航空航天大学 宇航学院, 北京 100191
  • 收稿日期:2010-01-14 出版日期:2011-04-29 发布日期:2011-04-29
  • 作者简介:王志飞(1981-),男,内蒙古鄂尔多斯人,博士生,wangfei54188@163.com.
  • 基金资助:

    航天创新基金资助项目(CASC0105)

Deflection prediction for inflatable wing based on artificial neural network

Wang Zhifei, Wang Hua, Jia Qinping, Han Jing   

  1. School of Astronautics, Beijing University of Aeronautics and Astronautics, Beijing 100191, China
  • Received:2010-01-14 Online:2011-04-29 Published:2011-04-29

摘要: 为实现对承载后柔性机翼挠度的准确预测,在全面分析柔性机翼挠度的影响因素基础上,应用正交试验法确定的影响柔性机翼挠度的主要因子作为输入变量,挠度作为输出变量,以大量试验数据为训练样本,通过多次试取隐含层和各隐含单元,并选取trainlm作为最优训练函数,最终建立了预测柔性机翼挠度的BP(Back Propagation)人工神经网络模型.在此基础上,随机选取试验结果中的12组试验样本,连续进行10次挠度预测,预测结果和试验实测值最大相对误差和标准方差分别为4.481%,1.033 7.解析结果表明:柔性机翼挠度预测结果与实验值吻合的较好,建立的人工神经网络预测模型具有较高的预测精度.

Abstract: To accurately predict the deflection of loaded inflatable wing, a basic impact of influence deflection was analyzed, method of orthogonal experiment was used to ascertain the main impact of influence deflection. The main impact of influence deflection was used as intputs and deflection was used as outputs. A BP artificial network model was established by using plenty of experimental statistics as training specimens, trying to access all kinds of crytic layers and elements, choosing trainlm as optimal function.Ten predictions were done continuously aiming at every group after twelve groups of specimens were selected from experimental results. The relative error between the predicted result and the experiment result is 4.48%, and standard deviation 1.033 7. The analysis results show that the rellative error between the predicted result and the messured reslut are slight for conrete specimens, which indicates that the established artificial network model has high prediction precison.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发