Water Resources and Power

There are many hydrological observation stations in the main stream of the Yangtze River, and the accurate medium long-term runoff forecasting is of great significance to the arrangement of power generation dispatching and the preparation of flood control and drought relief plan. This paper combined machine learning support vector machine(SVM) and long short-term memory neural network(LSTM) to predict the medium and long-term runoff of Pingshan, Yichang and Cuntan hydrological stations of the Yangtze River. The prediction results were evaluated by root mean square error, mean absolute percentage error and deterministic coefficient. In addition, support vector machine model and long short-term memory neural network model were used to forecast the hydrological situation of Yichang section monthly in 2021. The results show that the long short-term memory neural network model performs better than the support vector machine model in runoff forecasting during training and testing periods. Among the three sections of Yichang, Pingshan and Cuntan, the prediction results of Yichang are better than those of the other two stations. The results of the monthly rolling forecasting show that 2021 is a normal year for Yichang section.