Skip to main content

Real-time identification of vehicle motion-modes using neural networks

Journal Article


Abstract


  • A four-wheel ground vehicle has three body-dominated motion-modes, that is, bounce, roll, and pitch motion-modes. Real-time identification of these motion-modes can make vehicle suspensions, in particular, active suspensions, target on the dominant motion-mode and apply appropriate control strategies to improve its performance with less power consumption. Recently, a motion-mode energy method (MEM) was developed to identify the vehicle body motion-modes. However, this method requires the measurement of full vehicle states and road inputs, which are not always available in practice. This paper proposes an alternative approach to identify vehicle primary motion-modes with acceptable accuracy by employing neural networks (NNs). The effectiveness of the trained NNs is verified on a 10-DOF full-car model under various types of excitation inputs. The results confirm that the proposed method is effective in determining vehicle primary motion-modes with comparable accuracy to the MEM method. Experimental data is further used to validate the proposed method.

Authors


  •   Wang, Lifu (external author)
  •   Zhang, Nong (external author)
  •   Du, Haiping

Publication Date


  • 2015

Citation


  • L. Wang, N. Zhang & H. Du, "Real-time identification of vehicle motion-modes using neural networks," Mechanical Systems and Signal Processing, vol. 50-51, pp. 632-645, 2015.

Scopus Eid


  • 2-s2.0-84905855859

Ro Metadata Url


  • http://ro.uow.edu.au/eispapers/3334

Has Global Citation Frequency


Number Of Pages


  • 13

Start Page


  • 632

End Page


  • 645

Volume


  • 50-51

Place Of Publication


  • United Kingdom

Abstract


  • A four-wheel ground vehicle has three body-dominated motion-modes, that is, bounce, roll, and pitch motion-modes. Real-time identification of these motion-modes can make vehicle suspensions, in particular, active suspensions, target on the dominant motion-mode and apply appropriate control strategies to improve its performance with less power consumption. Recently, a motion-mode energy method (MEM) was developed to identify the vehicle body motion-modes. However, this method requires the measurement of full vehicle states and road inputs, which are not always available in practice. This paper proposes an alternative approach to identify vehicle primary motion-modes with acceptable accuracy by employing neural networks (NNs). The effectiveness of the trained NNs is verified on a 10-DOF full-car model under various types of excitation inputs. The results confirm that the proposed method is effective in determining vehicle primary motion-modes with comparable accuracy to the MEM method. Experimental data is further used to validate the proposed method.

Authors


  •   Wang, Lifu (external author)
  •   Zhang, Nong (external author)
  •   Du, Haiping

Publication Date


  • 2015

Citation


  • L. Wang, N. Zhang & H. Du, "Real-time identification of vehicle motion-modes using neural networks," Mechanical Systems and Signal Processing, vol. 50-51, pp. 632-645, 2015.

Scopus Eid


  • 2-s2.0-84905855859

Ro Metadata Url


  • http://ro.uow.edu.au/eispapers/3334

Has Global Citation Frequency


Number Of Pages


  • 13

Start Page


  • 632

End Page


  • 645

Volume


  • 50-51

Place Of Publication


  • United Kingdom