Monthly Runoff Interval Prediction Based on Fuzzy Information Granulation and Improved Neural Network

High-precision monthly runoff prediction results are of great significance to regional water resource management. However, with the changes in human activity, climate, and underlying surface conditions, sequence presents highly nonlinear random characteristics. In order improve accuracy prediction, this study proposed a model based on fuzzy information granulation (FIG) back propagation neural network (BPNN) improved genetic algorithm (FIG-GA-BP). First, FIG was used process original data generate three sequences minimum, average, maximum that can reflect rule changes. Then, algorithms (GA) were obtain optimal initial weights thresholds BPNN through selection, crossover, mutation. Finally, predict generated separately interval. The applied interval Linjiacun Weijiabu hydrological stations main stream Wei River Zhangjiashan station Jing River, tributary River. Compared FIG-BP, FIG-WNN, traditional BP model. show FIG-GA-BP had good effect, higher narrower range intervals. Therefore, has superiority practicability prediction.