Heuristic Methods for Reservoir Monthly Inflow Forecasting: A Case Study of Xinfengjiang Reservoir in Pearl River, China

摘要: Reservoir monthly inflow is rather important for the security of long-term reservoir operation and water resource management. The main goal present research to develop forecasting models inflow. In this paper, artificial neural networks (ANN) support vector machine (SVM) are two basic heuristic methods, genetic algorithm (GA) employed choose parameters SVM. When data series, both approaches inclined acquire relatively poor performances. Thus, based on thought refined prediction by model combination, a hybrid method involving two-stage process proposed improve forecast accuracy. method, ANN SVM are, first, respectively implemented data. Then, processed predictive values selected as input variables newly-built forecasting. Three models, ANN, SVM, developed in Xinfengjiang with 71-year discharges from 1944 2014. comparison results reveal that three have satisfactory performances prediction, performs better than terms five statistical indicators. an efficient tool dispatching reservoir.