A neural network approach for prediction of tyre rolling noise during indoor tests

Nowadays, noise has become one of the main concerns of environmental pollution, especially in the automotive and transport industry. Tyre noise has a core influence and manufacturers, pushed by regulations, invest several resources in experimental tests and numerical simulations to improve their products performances. In this context, predictive models represent useful tools to support the virtual prototyping. In this paper a statistical modelling approach for the prediction of sound intensity measurements in indoor condition is proposed. Particular attention is paid to the selection of the parameters related to tyre/road noise and to their processing. At first, sound intensity levels are measured for a set of tyres rolling on drum at different speeds. Then, noise-related tyre features are measured and processed. Eventually, an artificial neural network is used to define the statical model relating these features with the measured sound intensities. Promising results support the proposed approach.