2D, 3D Noise Modelling on Mobile GIS Application Through Machine Learning Based Models
- Publication Type:
- Thesis
- Issue Date:
- 2023
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
Vehicular traffic is one of the most significant engines for economic growth, therefore, its importance cannot be over emphasised. This noise causes a serious negative impact on the people living around the environment. Road traffic is known to be the major source of noise which often causes annoyance and interference. The aim of this research is to produce 2D and 3D noise maps for the study area and generate noise maps with the aid of mobile application through building two models (2D and 3D noise models) would be developed using machine learning algorithms, ArcGIS software, noise level, LiDAR data and road geometry and surrounding environments. So, the specific objectives and contributions of this research are developing noise sampling methods and generating observation points for modelling, model traffic noise in 2D and 3D using landuse regression (LUR) and machine learning methods, improve efficiency and scalability of noise models through integration and optimisation and develop noise visualisation tool for mobile application based on the models developed in this research. In this research was built two models: 2D noise model for roads and 3D noise model for buildings by using fewer noise samples. The 2D and 3D noise models were combined to produce 3D noise map for the study area. Also, the proposed models based on machine learning such as artificial neural network model (ANN), random forest (RF) and support vector machine (SVM), and the performance of three models were ascertained by calculating three performance measures: correlation (R), correlation coefficient (R2) and root mean square error (RMSE). The result of training and testing indicated ANN as the best model. The random forest (RF) has proven to be better than the support vector machine (SVM); the RMSE of the RF is less than RMSE of SVM. The RMSE of RF model showed (1.82, 6.00) and (9.83, 4.50) for training and testing 2D and 3D model respectively. While RMSE of SVM model was recorded to be (3.60, 6.16) and (10.34, 4.75) for training and testing 2D and 3D model, respectively. The main contributions of this study were the development of the noise prediction and propagation modelling methods. Both noise prediction and propagation models are valuable tools for traffic noise assessment during the highway design stage and to evaluate the impacts of traffic noise emitted from a vehicle on highways on the population.
Please use this identifier to cite or link to this item: