首页> 外文期刊>Institution of engineers journal (Inida) >Aircraft Parameter Estimation using Neural Network
【24h】

Aircraft Parameter Estimation using Neural Network

机译:基于神经网络的飞机参数估计

获取原文
获取原文并翻译 | 示例
           

摘要

The classical approach to estimate aerodynamic derivatives is applied using the techniques like output error, equation error and filter error methods. The output error and the filter methods may diverge if the initial estimates of these derivatives are poor. As an alternative approach, delta method applied to the trained neural network was investigated in the past. In this paper, yet another approach based on partial differential of the neural output is suggested for estimating the lateral stability and control derivatives and is compared against the delta method. It also compares the gradient descent and the scaled conjugate gradient algorithms to train the neural network. The proposed partial differential approach is one shot method, unlike the delta method and does not require initial estimates of the stability and control derivatives and is free from divergence problem. The results from the equation error method are used to substantiate the results obtained.
机译:使用诸如输出误差,方程误差和滤波器误差方法之类的技术来应用估算空气动力学导数的经典方法。如果这些导数的初始估计很差,则输出误差和滤波方法可能会有所不同。作为一种替代方法,过去已经研究了将三角洲方法应用于训练后的神经网络。在本文中,提出了另一种基于神经输出偏微分的方法来估计侧向稳定性和控制导数,并将其与增量法进行了比较。它还比较了梯度下降和比例共轭梯度算法来训练神经网络。所提出的偏微分方法是一种散弹方法,与增量方法不同,它不需要对稳定性和控制导数进行初始估计,并且没有发散问题。方程误差法的结果用于证实所获得的结果。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号