Urban transportation: Predicting acoustic and atmospheric pollution by means on neural networks.

Abstract

The prediction of the atmospheric pollution produced by transportation in urban areas is one of the main issues in which researchers are engaged worldwide since a long time. A lot of models have been introduced, ranging from simplified ones (that utilize literature data and phenomenological approaches), to very sophisticated methods (referring to complex mathematical techniques for solving the fluid dynamics algorithms). As matter of facts, passengers and freight vehicles are also responsible of acoustic pollution, that is particularly impacting in urban contexts. What it is new in this work is the consideration that both pollution phenomena occur in the same time and exert a double pressure on the urban environment. As that, any intervention aimed at the reduction of one of these problems, in the same time produces a positive effect on the other one environmental issue. Starting from a previous predicting model, introduced by authors, that is based on the assessment of a suitable neural network for analyzing the acoustic pollution produced by the urban transportation systems, a new neural model is here presented that is able to take contemporary into account atmospheric and acoustic releases from private and public transportation vehicles. The main problems in assessing such model are argued in the paper, along with the description of the structure of the neural network. The needed learning process of the neural network is also discussed and the main elements for a validation of the model are also singled out. The method is preliminary applied to the urban context of a south Italian medium size town.