Title: Neural network strategy for driving behaviour and driving cycle classification
Authors: Warren Vaz; Robert G. Landers; Umit O. Koylu
Addresses: Department of Mechanical & Aerospace Engineering, Missouri University of Science and Technology, 400 W. 13th Street, Rolla, Missouri 65409, USA ' Department of Mechanical & Aerospace Engineering, Missouri University of Science and Technology, 400 W. 13th Street, Rolla, Missouri 65409, USA ' Department of Mechanical & Aerospace Engineering, Missouri University of Science and Technology, 400 W. 13th Street, Rolla, Missouri 65409, USA
Abstract: The driving behaviour and the driving cycle type affect the range of an electric vehicle. A novel strategy that classifies driving behaviour as aggressive or defensive and driving cycles as highway or urban using accelerator and brake positions is proposed. A method to simulate aggressive and defensive driving behaviour using neural networks was developed and implemented. The neural network successfully differentiated between aggressive and defensive driving behaviour and highway and urban driving cycles in all 11 training cases. Furthermore, the neural network was able to properly classify the driving behaviour and the driving cycle type for four new driving cycles as well. The proposed method of classifying driving behaviour and driving cycles overcomes the limitations posed by identifying driving cycles. It provides real-time information about the driving behaviour and the driving cycle and is not limited to any particular driving cycle or group or driving cycles.
Keywords: driving behaviour classification; driving cycle classification; driving cycles; electric vehicles; EVs; neural networks; pattern recognition; supervised training; aggressive driving; defensive driving; highway driving; urban driving; accelerator positions; brake positions.
DOI: 10.1504/IJEHV.2014.065729
International Journal of Electric and Hybrid Vehicles, 2014 Vol.6 No.3, pp.255 - 275
Received: 24 Jan 2014
Accepted: 09 Jul 2014
Published online: 11 Nov 2014 *