A NEURAL-NETWORK APPROACH TO PREDICTION OF VEHICLE DRIVING COMFORT

We propose a novel approach to the prediction of vehicle driving comfo rt that allows the comfort issue to be integrated into the design proc ess. Whereas current practice relies on building prototype driving con figurations and asking a number of subjects to assess comfort, our app roach will enable the design engineer to obtain predicted comfort rati ngs on the basis of readily available design parameters, thus avoiding the expensive and time-consuming step of building a prototype for eve ry proposed design. We describe our approach, which focuses on the con tribution to comfort from the design parameters of the entire driving configuration, and the results of using a neural network to construct a relationship between these parameters and our empirical findings on subject comfort.