Search Results

This research proposes the use of Artificial Neural Networks (ANN) to predict the road input for road load data generation for variants of a vehicle as vehicle parameters are modified. This is important to the design engineers while the vehicle variant is still in the initial stages of development, hence no prototypes are available and accurate proving ground data acquisition is not possible. ANNs are, with adequate training, capable of representing the complex relationships between inputs and outputs. This research explores the implementation of the ANN to predict road input for vehicle variants using a quarter vehicle test rig. The training and testing data for this research are collected from a validated quarter vehicle model.

Sloped medians provide a run-off area for errant vehicles so that they can be safely stopped off-road with or without barriers placed in the sloped median. However, in order to optimize the design of sloped medians and the containment barriers, it is essential to accurately model the behavior of vehicles on such sloped terrain surfaces. In this study, models of a vehicle fleet comprising a small sedan and a pickup truck and sloped terrain surface are developed in CarSim™ to simulate errant vehicle behavior on sloped median. Full-scale crash tests were conducted using the vehicle fleet driven across a 9.754 meters wide median with a 6:1 slope at speeds ranging from 30 to 70 km/h. Measured data such as the lateral accelerations of the vehicle as well as chassis rotations (roll and pitch) were synchronized with the vehicle motion obtained from the video data.