Development Of Construction Noise Prediction Method Using Linear Support Vector Regression Model

Gam, Li Juen (2021) Development Of Construction Noise Prediction Method Using Linear Support Vector Regression Model. Final Year Project, UTAR.

PDF Restricted to Repository staff only until 31 December 2023. Download (3440Kb)

Abstract

Noise is an undesired and unpleasant sound that leads to adverse health effects on humans. Problem regarding occupational noise exposure is significantly increased, particularly in the construction industry. Over the years, construction noise caused a lot of noise-related health problems to the workers. Lack of research study in this field and a reliable method in noise prediction become the central problem in noise controlling and monitoring. Despite this, Haron, et al. (2012) had developed a simple prediction chart for noise prediction. However, there is some limitation on this model. For instance, the simple prediction chart does not consider the duty cycle of machines and workers. Other than that, the simple prediction chart only considers four types of angles from the noise source to receiver. Thus, this research project proposed to develop a noise prediction model to overcome the limitation stated previously by implementing linear support vector regression. As Google Colab provides a very stable platform for establishing a machine learning model, thus this platform was chosen to develop the noise prediction model. Moreover, python, the programming language, had been adopted in this study, whereas it is sufficient to handle an efficacious data structure. A total of seven noise prediction models had been established for different site aspect ratios, including 1:1, 1:2, 2:1, 1:4, 4:1, 1:8, and 8:1. In order to obtain the most optimized model, the parameters inside the model had been adjusted accordingly. By tuning the C and

Actions (login required)