ARTIFICIAL NEURAL NETWORK IN TRANSPORT RELATED POLLUTION MODELLING

Abstract

The healthy economic growth in India over the past few years has resulted in rapid urbanisation. The number of vehicles on roads is increasing exponentially especially in urban areas. This is leading to pollution in terms of both noise and air pollution. Traffic is the major contributor to both these types of pollutions in urban areas. Traffic accounts for 65% of air pollution and 55% of noise pollution in urban areas. In this dissertation, an attempt has been made to model traffic noise and air pollution, specifically CO concentration using neural networks. The prediction was done for the city of Delhi. Neural Networks were also used innovatively to design noise barrier. In case of noise prediction, the data was divided into four zones i.e. Silence, Residential, Commercial and Heavy Traffic Zone. For each zone, two scenarios were considered. First was with only classified traffic volume as input and the second with both classified traffic volume and classified traffic speed as inputs. The results showed that the noise level generated by neural networks was comparable to the observed levels. In case of CO concentration, four scenarios were considered. First, with only classified traffic volume input and the second with both classified traffic volume and meteorological variables. The third was with classified traffic volume, meteorological inputs and classified traffic speed inputs and the fourth was same as third minus the wind direction input. The results showed that CO concentration can be predicted satisfactorily using neural networks. Artificial Neural Networks were also used for design of noise barrier. The results show that given the traffic volume and speed data in a location and the required Leq, the height of the barrier required can be designed with good accuracy