A basic neural traffic noise prediction model for Tehran’s roads

We present an artificial neural network model to predict hourly A-weighted equivalent sound pressure levels (L_Aeq,1h) for roads in Tehran at distances less than 4 m from the nearside carriageway edge. Our model uses the UK Calculation of Road Traffic Noise (CORTN) approach. Data were obtained from 50 sampling locations near five roads in Tehran at nearside carriageway edge distances of less than 4 m. The data were randomly assigned to training, testing, and holdout subsets. Model training was carried out using the training and testing subsets and comprised 60% and 20% of the data, respectively. Model validation was performed using the remaining 20% of data as a holdout subset. We examine the overall model efficiency using non-parametric tests, such as the Wilcoxon matched-pairs signed-rank test for the training step and the Kolmogorov–Smirnov test for two independent samples for the validation step. Our results indicate that a neural network approach can be applied for traffic noise prediction in Tehran in a statistically sound manner. The Wilcoxon matched-pairs signed-ranks test detects no significant difference between the absolute testing set errors of the developed neural network and a calibrated version of the CORTN model.