A traffic noise prediction method based on vehicle composition using genetic algorithms

The effect of traffic composition on the noise pollution has been investigated in a small city Erzurum, located in eastern Turkey, where the population is about 400,000. Manual noise measurements and vehicle counts were performed at the four heaviest traffic points in the city for a period of 12 h. Using the information on vehicle composition and the maximum legal noise emissions of each type of vehicles, two prediction models based on genetic algorithms have been developed that can be used as tools for in-city traffic flow redesign. The models have been validated against some of the noise data. Measured traffic noise levels have been compared with predictions using the new models and some existing ones and a relatively good agreement has been obtained between them.