Research of dynamic load identification based on extreme learning machine

  • Authors:

  • Wentao Mao;Mei Tian;Guirong Yan;Xianfang Wang

  • Affiliations:

  • College of Computer and Information Technology, Henan Normal University, Xinxiang City, China,State Key Laboratory for Strength and Vibration, Xi'an Jiaotong University, China;Management Institute, Xinxiang Medical University, China;State Key Laboratory for Strength and Vibration, Xi'an Jiaotong University, China;College of Computer and Information Technology, Henan Normal University, Xinxiang City, China

  • Venue:
  • ISNN'12 Proceedings of the 9th international conference on Advances in Neural Networks - Volume Part I
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, the problem of multiple-input multiple-output dynamic load identification is addressed. First, from dynamic theory point of view, load identification is proved as a nonlinear multiple-input multiple-output regression problem which can be solved directly in black-box modeling manner from perspective of data analysis. Second, considering the good effect on multiple-input multiple-output problem, a recently proposed excellent machine learning algorithm, referred to as extreme learning machine, is introduced. Finally, a new identification method based on extreme learning machine is proposed to improve identification performance. Experiments on cylinder stochastic vibration system are conducted, demonstrating comparable results and encouraging performance of the proposed method compared with support vector machine based method in terms of identification accuracy, computational cost and numerical stability. A conclusion can also be drawn that extreme learning machine is better applicable to small-sample multiple-input multiple-output problem than support vector machine because of its ability to discover the dependencies among all outputs.