Applicability of modern correlation tools for ride comfort evaluation and estimation

The automotive world is currently shifting focus towards electric vehicles (EVs) and the market of
connected, autonomous vehicles (CAVs) is steadily growing. Vehicle ride comfort is an attribute which for
years now have been a factor which has a significant influence on vehicle development programmes. Due to
the complexity of ride comfort, achieving a good correlation between measured data and perceived comfort is
a challenging task. Creating well-handling vehicles with pleasant ride characteristics is becoming not enough,
as nowadays customers expect bespoke, tailored solutions such as active suspension systems instead of more
traditional, passive solutions. The presented study aims to analyse the usability of modern correlation tools,
such as artificial neural networks for objective and subjective data correlation, evaluation and explore the possibility of prediction of subjective responses based on the measured data. Data for the study was gathered on
the HORIBA MIRA proving ground and public roads. Measured parameters consisted of the vehicle accelerations, anthropometric data of the experiment participants and subjective evaluations of perceived vibration magnitudes. Subjective responses were gathered using a group of 22 participants. The obtained dataset was divided
into training and validation sets in the ratio of 80/20. Collected data was used in a correlation study using
artificial neural networks (ANNs). The created model achieved a high correlation level of R=0.91. Presented
study proves that correct use of advanced correlation techniques utilising artificial neural networks can create
comfort models allowing for subjective comfort response estimation. Such an approach could significantly reduce the time required for the vehicle development process and would allow for more comfortable, bespoke
vehicles in the future.