Continuous noise mapping predication techniques using the stochastic modelling

Abstract

A strategic noise map provides important information for noise impact assessment. However, current practices still use the unstandardised way which produces unreliable information for noise exposure monitoring. This research aims to develop new noise mapping prediction technologies in order to enhance the current noise prediction method and noise monitoring practices. The research work was divided into preliminary and primary studies. In the preliminary study, a survey was conducted to investigate current noise exposure problems among Malaysian industries. Questionnaires were designed based on the proposed theoretical framework and distributed to 215 respondents from six workplaces with different industrial background. The finding shows that only 10.7 % of respondents wear hearing protectors regularly, thus implies a high risk of noise exposure problems in these industries. Based on the results of the Chi-square test, the utilisation rate of hearing protectors was not affected by noise awareness and training factors, but it could be increased through supervision and provision of safety information. The primary research study proposed two prediction methods, namely a noise prediction chart and stochastic modelling to be used in the development of both the automation and stochastic simulation frameworks. An automation framework is a system that automatically refers to a noise prediction chart in predicting the noise levels at receiving points. A stochastic simulation framework incorporates a random walk process and Monte Carlo approach to simulate movement and noise emission levels of machinery in a defined mapping area. Two prototyping softwares, namely Prototype I and Prototype II, were programmed using the MATLAB programming software in order to facilitate each proposed framework. Both prototyping software generated outputs such as strategic noise map, noise risk zone, and noise information. For software validation, a comparison of prediction and measurement results from case studies was performed. Eight case studies of field measurements from different industries were used to obtain the prediction inputs and noise levels from control points. The absolute differences between prediction and measurement values at the control points were computed to determine the accuracy of prediction results for each prototype. In general, the prediction results of Prototype I and II had a good agreement (≤ 3 dBA) with the results obtained from measurement for most of the case studies. Both prototypes could reflect the complex and dynamic noise circumstances in a workplace. This study has significantly in advanced noise mapping prediction technologies and the prototypes produced could be beneficial as new noise monitoring tools in current industrial practices.