Design of Optimal Noise Barrier for Metropolitan Cities Using Artificial Neural Networks

The study compares the available models that describes the insertion loss of noise barrier and also describes the development of the ANN model for Indian traffic scenario. The height of the noise barrier was made to increase with an increment of 0.1 m from a height of 2 to 15 m to carefully examine the behaviour of the curves obtained from both models. The output from the best model based on the comparisons was then selected for the design of the noise barrier. Sri Aurobindo Marg (near AIIMS, New Delhi) was selected as the site for assessing the traffic flow parameters such as the traffic flow, average speed and noise levels at the site. These parameters together with the insertion loss results served as the input to the desired ANN model. The ANN model with various architectures was trained with 70% of the data, tested with 15% of data and validated with 15% of the data, while the root mean square error and the corresponding correlation coefficients were the judging parameters for the ANN model.