A stochastic and microscopic model to predict road traffic noise by random generation of single vehicles' speeds

Road traffic noise is the major component of acoustic environmental pollution both in urban and rural areas. For this reason, much effort has been put into developing models to assess its impact. However, literature models are often suitable for standard conditions but can fail in non-standard ones, i.e., when the single vehicle speed cannot be neglected. Moreover, input data to literature models are not always available, e.g., if the road infrastructure is still in the design phase. The presented approach aims to try to overcome these shortcomings using a microscopic and stochastic-core model, in which the speed of each vehicle can be randomly generated using a specific speed distribution. The validation of the model, investigated through a statistical analysis of simulated continuous equivalent sound pressure levels, the error distribution, and the calculation of commonly used error metrics suggests that the proposed methodology provides good estimations of traffic noise. The errors of the model computed as the differences between measured and simulated sound levels, can be described as a distribution curve with a -0.6 dBA mean and a standard deviation of 2.3 dBA. The error metrics confirm the model's goodness, with a mean absolute error of 1.84 dBA and a coefficient of variation error of 0.03.