PGS. TS. Ngô Văn Quận[1], TS. Đỗ Phương Thảo1, ThS. NCS. Nguyễn Xuân Thịnh[2], ThS. Nguyễn Văn Tiến[3]
Abstraction: Forecast of the water flow to the reservoir has a very important significance in the operation of the reservoir safety and rational distribution of water for the water use needs. Consequently, there have been many studies enhance the reliability of the forecast of the amount of water to serve for the management, operation of the reservoir. Therefore, main objective of study is to apply of artificial neural network (ANN) direct use of reverse propagation algorithm to forecast the water flow to the Của Đạt reservoir before 3 days on the basis of the analysis, assessment for two scenarios: (1) To forecast water flow to reservoir only consider about flow element and (2) consider both of flow and rainfall elements. The results showed that, when considering both of flow and rainfall factors for higher accuracy. In particular, with the data of the Cửa Đạt station is error (RMSE) reduced from 75.04 to 56.35 m3/s m3/s and the index of R2increased from 0.68 up 0.72, with similar error (RMSE), Cam Thủy station also dropped and index R2 increased. This research results provide a useful method in forecasting, that is particularly meaningful in the management of water resources and significant contributions in building the reservoir operation management scenarios are more active.
Attached link here
Submitted manuscript paper_ANN-gui dang