期刊文献+

基于SFOS-ELM的电动无人机动力套装在线建模

On-Line Modeling of Electric Drone Power Set Based on SFOS-ELM
下载PDF
导出
摘要 传统超限学习机(ELM)梯度下降算法需要多次迭代求解,并沿用批量式学习模式导致必须一次性输入所有训练样本,计算量大耗时多,难以应用于低成本微小型电动无人机动力套装的在线建模。为解决上述问题,研究了一种基于平滑融合型在线序列超限学习机神经网络(SFOS-ELM),采用增量式学习模式将分批处理训练样本和逐次迭代相结合,用于微小型电动无人机动力套装在线建模。所提方法训练速度快、泛化性能强,能够实时跟踪无人机动力套装的动态特性。经数学仿真和实际动力套装数据验证,所提方法在线建模速度快、适应能力强,可以实现微小型电动无人机在线建模。 The traditional extreme learning machine(ELM)gradient descent algorithm requires multiple iterative solutions,and its batch learning mode needs inputting all training samples at one time,which results in the increased computational complexity and time consuming,so it is difficult to apply to the online modeling of low-cost micro-electric drone power set.A smooth fusion type online sequence extreme learning machine neural network(SFOS-ELM)for online modeling of electric drone power set was presented in this study,which combines batch processing training samples with successive iterations using incremental learning mode.The SFOS-ELM method presented is with more fast training speed,more strong generalization performance,and the dynamics of the power set of the drone can be tracked in real time.The mathematical simulation and actual power set try were given.The method presented shows fast online-modeling performance with strong adaptive ability,which is competent in online modeling of electric drone power set.
作者 赵丙将 张玉民 黄文晔 ZHAO Bing-jiang;ZHANG Yu-min;HUANG Wen-ye(Beihang University,Beijing 100191,China)
出处 《计算机仿真》 北大核心 2020年第8期44-48,62,共6页 Computer Simulation
基金 国家自然科学基金(61374131)的资助
关键词 超限学习机 无人机 动力套装 建模 Extreme learning machine(ELM) Drone power set Model
  • 相关文献

参考文献3

二级参考文献81

  • 1房建成.永磁无刷直流电机控制技术与应用[M]. 北京:机械工血出版社,2011.
  • 2Thomas Salem,Tim A Haskew. Simulation of the brushlessDC machine [ C ] // Proceedings of the Twenty-SeventhSoutheastern Symposium on System Theory ( SSST,95 ).USA:[s. n. ].1995:18 -22.
  • 3Ionel D M,Eastham J F,Betzer T. Finite element analysis ofa novel brushless DC motor with flux barriers [ J ]. IEEETrans, on Magnetics ( S0018 — 9464 ),1995,31 (6 ) : 3749-3751.
  • 4Jeon Y S,Mok H S,Choe G H,et al. A new simulation mod-el of BLDC motor with real back EMF waveform [ C ] //Proceeding from Computers in Power Electronics. The 7thWorkshop on Volume Computers in Power Electronics.USA: [s. n. ].2000:217 - 220.
  • 5Rumelhart D E, Hinton G E, Williams R J. Learning representations by back-propagating errors. Nature, 1986, 323: 533-536.
  • 6Hagan M T, Menhaj M B. Training feedforward networks with the marquardt algorithm. IEEE Trans Neural Netw, 1994, 5:989-993.
  • 7Wilamowski B M, Yu H. Neural network learning without backpropagation. IEEE Trans Neural Netw, 2010, 21: 1793-1803.
  • 8Chen S, Cowan C, Grant P. Orthogonal least squares learning algorithm for radial basis function networks. IEEE Trans Neural Netw, 1991, 2:302-309.
  • 9Li K, Peng J X, Irwin G W. A fast nonlinear model identification method. IEEE Trans Automat Contr, 2005, 50: 1211-1216.
  • 10Hornik K. Approximation capabilities of multilayer feedforward networks. Neural netw, 1991, 4:251 257.

共引文献85

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部