File(s) not publicly available
Narx Neural Network Modelling of Hydraulic Suspension Dampers for Steady-State and Variable Temperature Operation
Two NARX-type neural networks are developed for modelling nonlinear dynamic characteristics of passive twin-tube hydraulic dampers used in vehicle suspension systems. Quasi-isothermal and variable temperature NARX models are rigorously tested and compared with a state-of-the-art physical model proposed by Duym and Reybrouck (1998) and Duym (2000). Measured damper data, generated under isothermal and temperature varying conditions, is used for NARX training, physical model calibration, and predictive comparisons. Test kinematics include high amplitude sinusoidal displacements up to 14 Hz, and realistic random road profiles. The NARX models are trained via 'teacher forcing' and the feedforward backpropagation algorithm using both 'Early Stopping' and Bayesian Regularisation. Stable network design is also examined using the minimum posterior prediction error as the criterion for selecting a good network from a small number of tests. Calibration of the physical model proves highly complicated owing to considerable nonlinearity-in-the-parameters, requiring use of Sequential Quadratic Programming with an implicitly nonlinear constraint. The paper shows that NARX neural network modelling is vastly superior in terms of calibration efficiency, and prediction times, whilst offering roughly similar, if not better, model accuracy.
History
Publication status
- Published
Journal
Vehicle System DynamicsISSN
0042-3114Publisher
Taylor & FrancisExternal DOI
Issue
5Volume
40Page range
283-328Pages
44.0Department affiliated with
- Engineering and Design Publications
Notes
In this paper two NARX neural network models are compared with a state-of-the-art physical model, all calibrated with realistic damper data. Patel proposed the NARX models (as part of her DPhil). These were initially trained using `Teacher Forcing, standard Back-Propagation, and `Early Stopping. Dunne improved the training strategy using Bayesian Regularisation, and a stable-network design methodology. The paper shows that the NARX models are vastly superior in terms of calibration and prediction efficiency, at the same time as offering similar model accuracy. This paper is of major importance to the design and tuning of vehicle suspension systems.Full text available
- No
Peer reviewed?
- Yes
Legacy Posted Date
2012-02-06Usage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC