Publication: Water Quality Forecasting in Shrimp Cultures based on Monte Carlo Tree Search
Issued Date
2021-04-01
Resource Type
Other identifier(s)
2-s2.0-85106632578
Rights
Mahidol University
Rights Holder(s)
SCOPUS
Bibliographic Citation
2021 7th International Conference on Engineering, Applied Sciences and Technology, ICEAST 2021 - Proceedings. (2021), 199-204
Suggested Citation
Dhanachai Pinitsava, Piyaporn Surinlert, Worapan Kusakunniran Water Quality Forecasting in Shrimp Cultures based on Monte Carlo Tree Search. 2021 7th International Conference on Engineering, Applied Sciences and Technology, ICEAST 2021 - Proceedings. (2021), 199-204. doi:10.1109/ICEAST52143.2021.9426250 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/76667
Research Projects
Organizational Units
Authors
Journal Issue
Thesis
Title
Water Quality Forecasting in Shrimp Cultures based on Monte Carlo Tree Search
Other Contributor(s)
Abstract
Shrimp is a sensitive creature and cannibalism. When shrimp die because of some factor in the water, for example, the temperature suddenly changed. Shrimp will start the cannibalization process. Farmers need to understand the ecological process to avoid this situation. Water management is one of them, and water quality properties take the main role of it. If farmers know the water's current and future situations, it could help them handle unexpected/unforeseen situations. In this research, forecasting water quality values by using Monte Carlo Tree Search (MCTS) is proposed. Salinity, pH, Dissolved Oxygen, Temperature was collected by IoT Arduino based device with Solar cell as a power source and sent data using NB-IoT module. Linear interpolation was manipulated raw data for creating a new dataset of every 30 minutes. The data was given a grade from 1 to 5. MCTS forecast value by cutting the outliner in the selection phase. After selecting the node, expand the selected node, simulate the node until found the target node, give a score, and calculate and update the score back to the beginning node. The result is the route from the beginning node to the target node that has the highest score. The device can float on the water and work all day all night. The data collected from the device does not cover the entire pond's water quality because there is one device in the large area of shrimp ponds. The MCTS can forecast the water quality in the small area around the device. When the farmer knows the water's future situation will help them reduce the risk of losing the shrimp.