Source localization for illegal plastic burning in Malaysia via CFD-ANN approach
Illegal plastic burning has caused several environmental and health impacts on society. It is important to locate the burning source quickly to mitigate the emission before people are exposed to the toxic gases. However, the conventional methods of source localization such as trained dogs, sensors,...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Published: |
Elsevier
2022
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/103256/ |
| _version_ | 1848863972231479296 |
|---|---|
| author | Yu, H. L. Chen, B. H. Kim, K. S. Siwayanan, P. Thomas Choong, S. Y. Ban, Z .H. |
| author_facet | Yu, H. L. Chen, B. H. Kim, K. S. Siwayanan, P. Thomas Choong, S. Y. Ban, Z .H. |
| author_sort | Yu, H. L. |
| building | UPM Institutional Repository |
| collection | Online Access |
| description | Illegal plastic burning has caused several environmental and health impacts on society. It is important to locate the burning source quickly to mitigate the emission before people are exposed to the toxic gases. However, the conventional methods of source localization such as trained dogs, sensors, and infrared camera are limited and less efficient. This research paper was conducted to study the combination of Computational Fluid Dynamics (CFD) and machine learning method on the plastic burning location assessment. 8 sensors were placed in a 530 m radius around the residential area in Telok Panglima Garang city in the computational domain to detect the concentration of the toxic gases released (methane and benzene) from 12 different possible illegal burning locations. A total of 65 training data sets and 7 validation sets under different burning locations, wind speeds, and wind directions were obtained using CFD approach. According to the simulation, it was found that the sensor readings vary under different atmospheric conditions. Besides, the wind direction and wind speed will affect the direction of gas dispersion and mixing effect, which results in different sensor values. The data sets obtained from the generated simulation were used for the machine learning process in the Artificial Neural Network (ANN) model to study the trend for each case. In this report, the ANN model includes 16 input, 4 hidden, and 12 output neurons. The model can achieve 85.71% validity with an average error of 3.86%. |
| first_indexed | 2025-11-15T13:41:24Z |
| format | Article |
| id | upm-103256 |
| institution | Universiti Putra Malaysia |
| institution_category | Local University |
| last_indexed | 2025-11-15T13:41:24Z |
| publishDate | 2022 |
| publisher | Elsevier |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | upm-1032562023-11-22T02:49:41Z http://psasir.upm.edu.my/id/eprint/103256/ Source localization for illegal plastic burning in Malaysia via CFD-ANN approach Yu, H. L. Chen, B. H. Kim, K. S. Siwayanan, P. Thomas Choong, S. Y. Ban, Z .H. Illegal plastic burning has caused several environmental and health impacts on society. It is important to locate the burning source quickly to mitigate the emission before people are exposed to the toxic gases. However, the conventional methods of source localization such as trained dogs, sensors, and infrared camera are limited and less efficient. This research paper was conducted to study the combination of Computational Fluid Dynamics (CFD) and machine learning method on the plastic burning location assessment. 8 sensors were placed in a 530 m radius around the residential area in Telok Panglima Garang city in the computational domain to detect the concentration of the toxic gases released (methane and benzene) from 12 different possible illegal burning locations. A total of 65 training data sets and 7 validation sets under different burning locations, wind speeds, and wind directions were obtained using CFD approach. According to the simulation, it was found that the sensor readings vary under different atmospheric conditions. Besides, the wind direction and wind speed will affect the direction of gas dispersion and mixing effect, which results in different sensor values. The data sets obtained from the generated simulation were used for the machine learning process in the Artificial Neural Network (ANN) model to study the trend for each case. In this report, the ANN model includes 16 input, 4 hidden, and 12 output neurons. The model can achieve 85.71% validity with an average error of 3.86%. Elsevier 2022 Article PeerReviewed Yu, H. L. and Chen, B. H. and Kim, K. S. and Siwayanan, P. and Thomas Choong, S. Y. and Ban, Z .H. (2022) Source localization for illegal plastic burning in Malaysia via CFD-ANN approach. Digital Chemical Engineering, 3. art. no. 100029. pp. 1-19. ISSN 1385-8947 https://www.sciencedirect.com/science/article/pii/S2772508122000205 10.1016/j.dche.2022.100029 |
| spellingShingle | Yu, H. L. Chen, B. H. Kim, K. S. Siwayanan, P. Thomas Choong, S. Y. Ban, Z .H. Source localization for illegal plastic burning in Malaysia via CFD-ANN approach |
| title | Source localization for illegal plastic burning in Malaysia via CFD-ANN approach |
| title_full | Source localization for illegal plastic burning in Malaysia via CFD-ANN approach |
| title_fullStr | Source localization for illegal plastic burning in Malaysia via CFD-ANN approach |
| title_full_unstemmed | Source localization for illegal plastic burning in Malaysia via CFD-ANN approach |
| title_short | Source localization for illegal plastic burning in Malaysia via CFD-ANN approach |
| title_sort | source localization for illegal plastic burning in malaysia via cfd-ann approach |
| url | http://psasir.upm.edu.my/id/eprint/103256/ http://psasir.upm.edu.my/id/eprint/103256/ http://psasir.upm.edu.my/id/eprint/103256/ |