Title: Artificial road input data synthesis: a full vehicle model case study
Authors: Adebola Ogunoiki; Oluremi Olatunbosun
Addresses: School of Mechanical Engineering, University of Birmingham, Birmingham, B15 2TT, UK ' School of Mechanical Engineering, University of Birmingham, Birmingham, B15 2TT, UK
Abstract: In order to reduce the time and cost of developing a vehicle, it is important that virtual durability testing is carried out. In this research project, the aim is to predict the road input for the virtual durability test simulation of a new vehicle variant by transforming the data from a predecessor model using the vehicle's configuration parameters to generate a new and representative road input. To achieve this, a full vehicle model of a sport utility vehicle (SUV) is developed and validated with test data collected on a proving ground; this model is used to generate data to train and validate a NARX based artificial neural network (ANN) tool which is then subsequently used to predict the road input to the new variant of the vehicle. The use of ANNs in this project shows one of the many potentials of artificial intelligence in developing virtual capabilities within the automotive industry.
Keywords: multi-body dynamics; ANN; artificial neural network; durability; CAE; computer aided engineering; RLD; road load data; full vehicle model; vehicle variant; QanTiM; SIMPACK; MBS; multi-body simulation.
International Journal of Vehicle Design, 2021 Vol.85 No.2/3/4, pp.197 - 229
Accepted: 10 Jul 2020
Published online: 19 Jan 2022 *