Buon Kuop Hydropower Company: Application of artificial intelligence to forecast floods early
With the goal of perfecting the flood warning system, Buon Kuop Hydropower Company (under the Power Generation Corporation No.3) has actively implemented digital transformation for hydrological data collection, flood warning and developing digital maps for downstream areas of Buon Tua Srah, Buon Kuop and Srepok 3 hydropower plants.
In 2020, the company has installed 12 additional rain gauge stations and 4 automatic water level monitoring stations, bringing the total number of stations to 23 rain gauge stations and 7 water level monitoring stations. Real data from monitoring stations are connected to the center and serve as a database to build a water flow forecasting model for the reservoirs of Buon Tua Srah, Buon Kuop and Srepok 3 Hydropower Plants.
It is special that this model is self-studied and developed by engineers of Buon Kuop Hydropower Company. Based on the collected data, the engineers have applied AI (artificial intelligence) to analyze, forecast and provide early warning.
So far, the model has obtained quite positive results. Through the process of comparing and cross-checking with the measured data and the actual situation, the water flow forecasting model for Buon Tua Srah reservoir is now about 75% accurate and becomes a useful channel for the company’s reference.
For downstream areas, flood discharging warning systems of Buon Kuop Hydropower Company have been operating since 2009 and gradually being switched to remote warning systems via mobile phone waves. Currently, 20 warning stations have been arranged along the riverbank in downstream areas, ensuring that notices about the plant operation status, notices on the regulation of water discharge of power plants reach the local authorities and local people in the region.
According to the plan, in 2021, Buon Kuop Hydropower Company will procure, self-study, design and install 06 additional rain monitoring stations, at the same time collect more rain forecast data from meteorological stations and other catchments, study and upgrade the Matlab’s Neural network application (nntool) to further improve the forecast accuracy.
In addition, the company will study and develop digital maps for downstream areas of power plants; building a new information website and coordinating with local authorities to install more remote warning stations, with focus ondensely populated areas, flooded areas, areas with many livestock and aquaculture households in downstream areas of power plants.
It is expected that Buon Kuop Hydropower Company can improve the forecast accuracy to about 90%.