APPLICATION OF RECURRENT NEURAL NETWORKS IN ASSESSING DRIVERS' OSCILLATORY COMFORT DURING FORE-AND-AFT VIBRATIONS

Abstract

Vibrations emanating from the environment subject vehicle occupants to conditions that significantly influence their comfort and safety. These vibrations, particularly those stemming from horizontal and vertical movements, are transmitted through the vehicle's structure, from the seat to the driver's body. This study focuses on the seat-to-head transmission response function, which encapsulates the relationship between vibrations at the seat/head interface and the resultant motion responses of the driver's head. To address this, an artificial neural network model was developed based on experimental measurements involving ten healthy female subjects who were exposed specifically to horizontal fore-and-aft vibrations. Training of the artificial neural network was conducted using values from the transfer functions derived from these experiments. The findings indicate that the devised model can accurately forecast transfer function values within the range of the experimental data upon varying the input parameters, thereby offering insights into managing oscillatory comfort for automobile drivers.