Prediction of traffic noise induced annoyance of vendors through noise indices using structural equation modeling: Artificial neural network model. Issue 2 (30th July 2022)

• Record Type: Journal Article
• Authors: Das, Chidananda Prasad; Goswami, Shreerup; Das, Mira
• Is Part Of: Environmental quality management. Volume 32:Issue 2(2023)
• Page Start: 29
• Page End: 44
• Publication Date: 2022-07-30
• Subjects: annoyance -- artificial neural network -- noise indices -- structural equation modeling -- transportation noise; Industries -- Environmental aspects -- Periodicals; Environmental protection -- Periodicals; Industrial management -- Periodicals; Environmental engineering -- Periodicals; 658.408
• DOI: 10.1002/tqem.21905
• Languages: English
• ISSNs: 1088-1913

Abstract

Abstract: Urbanization and population growth, as well as the movement of diverse vehicles on urban roadways, all contribute to excessive noise in the urban acoustic environment. Annoyance, cardiovascular disease, dementia, hypertension, stress, sleeping problem, irritation, hair‐fall, headache, are the most common reported problems due to traffic noise exposure. The "Structural Equation Modeling–Artificial Neural Network Model" was used in the current study to predict traffic noise induced annoyance among 100 vendors whose shops are located from Khandagiri to Baramunda along the National Highway 16 of the smart city Bhubaneswar Odisha. Due to the movement of large number of heavy vehicles as well as medium and light weight vehicles, these roadways remain noisy throughout the day. Moreover, the annoyance questionnaire was developed in compliance with ISO/TS 15666 criteria for assessing annoyance level. The combination of SEM and ANN is rarely seen in acoustics, especially in noise studies. However, in this study, both SEM and ANN are used to predict annoyance using various noise indices. Moreover, the SEM model revealed a significant association between "Equivalent Noise Level" (Leq) and annoyance ( p ‐value = 0.031), "Minimum Noise Level" (LMin) and annoyance ( p ‐value = 0.049), "Background Noise Level" (L90 ) and annoyance ( p ‐value = 0.047), "Noise Pollution Level" (NPL) and annoyance ( p ‐value = 0.038). These associations indicate that Leq, LMin, L90 and NPL have a significant effect on annoyance. Furthermore, the PLS algorithm output from the measurement model verified a 47 percent variance in annoyance level. The ANN model confirmed that NPL is the most significant predictor of noise annoyance, followed by LMin, Leq, and L90 . Moreover, the ANN model can predict annoyance with an accuracy of 68.5 percent.