Salta al contenuto principale
Passa alla visualizzazione normale.

MARIA LA GENNUSA

Forecasting noise levels by means of neural networks for assessing urban traffic policies

Abstract

Within the assessment of the sustainability of plans and actions concerning the built environment, the transportation sector plays an increasing role, due to its importance in the economic and social life of countries. As that, the analysis of the sustainability concerning the transportation sector is now often embodied into the so called Strategic Environmental Analyses (SEA), that should provide local administrators with easy criteria for ranking the environmental suitability of designing policies, and that would seem to encounter the needed features for a correct evaluation of the urban masterplan. The urban noise forecast is very useful for local administrations, which have presently to cope with the urgent problem of assessing the environmental sustainability of policies in urban contexts. Several methods and indicators have been introduced, in order of singling out the environmental impact of the transport sector but, nevertheless, local administrators are called to manage complex set of data. In this work a software based on neural networks is proposed with the aim to forecast the levels of noise pollution. The proposed method essentially allows to link some typical urban planning data with the traffic noise, and to assess the influence of the program choices regarding the noise levels. So, it is possible to forecast new scenarios of acoustic pollution in urban contexts, and to compare the obtained data with the limits of law. To overcome the problem due to the different types of parameters which influence the acoustic pollution phenomena, a neural network has been used.