A Multiple Linear Regression Approach in Modeling Traffic Noise

An analytical model is developed to predict road traffic noise for busy roads of Delhi, India. Equivalent continuous sound pressure level, L AeqT , is analyzed at eight different busy road locations of Delhi. A multiple linear regression analysis is conducted to predict the single noise metrics L Aeq in terms of traffic flow rate (Q), percentage of heavy vehicles (H), and average traffic speeds (V). The model so developed is validated with actual experimental data. The coefficient of regression for test data set is observed to be 0.74 between predicted and experimental L Aeq values. The work thus shows that a validated analytical model can be useful for predicting noise levels and conducting the noise impact assessment studies in Delhi. The accuracy of model so developed can be further enhanced by feeding more real-time data from different locations with varied traffic density.