Firstly, a combined weighting prediction model of monthly runoff was presented on the basis of systematical research on the method of stationary time series, BP artificial neural network and periodic analysis. Secondly, after studying the risk rate calculating method of reservoir dispatch, the risk of reservoir impounding and power generation dispatch for runoff with and without forecast was calculated using the methods of random simulating and maximum entropy respectively. Finally, Danjiangkou reservoir was taken as an example to validate this model. The results show that the model can predict the monthly runoff series and give a more accurate prediction than traditional methods, thus provide a better way for the medium and long term runoff forecasting. At the same time, the risk information of Nalan reservoir calculated by this method can provide a decision-making basis for decision maker to dispatch the time of reservoir impounding and the runoff for power generation which can ensure the reservoir operating in a better way.
|