Traffic noise recordings have been analysed and it has been established experimentally that observations presented in the form of graphs approach a Gaussian distribution. To account for this Gaussian distribution, a statistical model of traffic noise has been evolved after which it is used with reasonable success as a prediction mechanism to evaluate another random noise situation similar to that of traffic noise. The model also enables experimental data relating to noise surveys to be characterised by a limited number of parameters capable to describe a particular random noise situation in a physically meaningful manner. The article includes various practical techniques used in the statistical analysis of traffic noise and a discussion to which the theoretical model is verified experimentally.
Abstract