Comparison of ANN and Analytical Models in Traffic Noise Modeling and Predictions

Abstract

This paper demonstrates the applications of artificial neural networks to predict the equivalent continuous sound level $$(L_\mathrm{Aeq})$$ and 10 Percentile exceeded sound level $$(L_\mathrm{10})$$ generated due to traffic noise for various locations in Delhi. A Model based on back-propagation neural network was trained, validated, and tested using the measured data. The work shows that the model is able to produce accurate predictions of hourly traffic noise levels. A comparative study shows that neural networks out-perform the multiple linear regression models developed in terms of total traffic flow and equivalent traffic flow. The prediction model proposed in the study may serve as a vital tool for traffic noise forecasting and noise abatement measures to be undertaken for Delhi city.