Flood Forecast In Douhe Reservoir

Inflow flood forecasting is the basis of reservoir flood control operation,the forecast period and the accuracy and reliability of flood forecasting have important influence on flood prevention and flood resource utilization.At present,there are some usual methods of flood forecasting.Firstly,according to the basin's natural and geographical conditions and field observation and investigation,research into the historical flood information and corresponding to the flood of hydrology,rainfall,natural geography information,through the establishment of flood and water,rainfall information or rate parameters of flood forecasting model,runoff forecast predicting flood;then,the same is based on the historical flood information and corresponding to the flood of hydrology,rainfall,natural geography information and analysis calculation unit hydrographs,runoff forecast,for calculating the flood hydrograph,so as to achieve the purpose of flood forecasting.That the research of Douhe reservoir basin is very small causes the lack of historical flood data.The usual flood forecast methods are not useful to Douhe reservoir prediction.According to the information condition,this paper established the flood forecasting model of Douhe reservoir mining technology based on modern data.The practical example shows that this model is simple and practical,and has a certain precision.The main work is as follows:(1)collation and analysis and examination for the "three nature" of these data which historical floods,rain and corresponding natural geography of Douhe Reservoir basin;(2)the application of genetic programming to establish P-Pa-R model of Douhe reservoir inflow flood;(3)the BP neural network is used to established rainfall-antecedent rainfall-flood peak flow forecast model,in order to overcome the shortcomings of traditional BP neural network initialization is easy to fall into local optimal value of the shortcomings,simulated annealing algorithm and genetic algorithm is introduced to the model parameter optimization and calibration,and achieved good results;(4)the flood forecasting accuracy of different algorithms are evaluated and compared.