A Data-based Water-inflow Forcasting Method for Small/medium Sized Hydropower Plants in Spot Market

Accurate prediction of water inflow is very important for a small and medium sized hydropower plant to operation and schedule. Therefore, this paper proposes a data-based approach to forecast the inflow of small and medium hydropower stations. An integrated neural network, which combines back-propagation (BP) neural network and Long-short-term memory (LSTM) neural network, used in the method proposed in this paper. Because the flood season and the non-flood season have different characteristics of the water inflow cycle, the training is carried out separately. Based on the historical measured inflow data of small and medium power stations, it is verified that the prediction accuracy of the proposed model in both flood season and non-flood season have reached Level A.