MCB_4_16.pdf (258.53 kB)

Structured learning of a non-linear dynamic system component for vehicle motion simulation

Download (258.53 kB)
conference contribution
posted on 27.04.2011 by Matt Best, Andrew P. Newton
Here we consider the design of a general subsystem model which is required to operate in series with the known dynamics of a plant or actuator, to achieve a desired overall system response. The example model is used in series with a motion platform to emulate vehicle handling dynamics. The method works in stages, first isolating the required linear response by fitting a frequency response function, then modelling this with a fixed order linear system in modal canonical form. A nonlinear saturation is then optimised for each modal state. The results are demonstrated for simulated and vehicle test data, and these achieve the principal objectives, of low state and parameter order. Some limitations to the method emerge – principally that there remain challenges to extension of the model to multi-input / output operation.

History

School

  • Aeronautical, Automotive, Chemical and Materials Engineering

Department

  • Aeronautical and Automotive Engineering

Citation

BEST, M.C. and NEWTON, A.P., 2006. Structured learning of a non-linear dynamic system component for vehicle motion simulation. Proceedings of the 8th International Symposium on Advanced Vehicle Control (AVEC), Taipei, Taiwan, 20th-24th August, pp. 559-564.

Publisher

© Society of Automotive Engineers of Japan (JSAE)

Version

AM (Accepted Manuscript)

Publication date

2006

Notes

This is a conference paper.

ISBN

109860059470

Language

en

Exports