Use of artificial neural networks to assess train horn noise at a railway level crossing in India

Abstract

Urban environment noise is a complex mixture of transportation, industrial, household, and recreational noise, which is identified as an emerging environmental threat. Present study monitors and evaluates a noise pollution hotspot: a railway level crossing, where several activities related to transportation noise were involved. Train honking, train movement, road vehicles, and pedestrians contribute to the noise level at a railway level crossing. Train horns are generally performed as train approach railway level crossings and they are mandatorily used to alert road users. However, the train horns are regarded as nuisance to the nearby residents. A detailed evaluation of train horn effectiveness is very much essential in the current contemporary environment. Thus, the main objective of this study is to measure noise levels emanating from train horns at a level crossing with due consideration to train types and climatic conditions. A comprehensive noise monitoring survey was conducted at an access-controlled level crossing. Furthermore, an artificial neural network (ANN)-based railway noise prediction model was developed to forecast maximum ([Formula: see text]) and equivalent (Leq) noise levels. Results revealed that train horn produced impulsive sound signals which fall under high frequency one-third octave bands causing severe irritation to trackside inhabitants. The proposed ANN models produced accurate results for [Formula: see text] and Leq noise levels and this model is identified as a vital tool for railway noise abatement. The results from this study are helpful to the urban planning and development authorities to implement strategic laws and policies to eradicate the urban environment noise.