| Researchers have utilized different algorithms to solve reservoir system optimization problems, and a large amount of research achievements have been made. However, there are still lots of problems to considered and explored in the population of their applications due to the uncertainties of inflow and its forecasts. This dissertation focuses on the use of Quantitative Precipitation Forecasts (QPFs) in medium-and-long-range runoff forecasting and hydropower operation of reservoirs system in Yalong river basin. Hydropower operation optimization is studied in depth using stochastic dynamic programming considering the uncertainty of inflow and forecasts. Then the primary achievements of this dissertation include as follows:(1) A new method for classification of long-range runoff is proposed by decision tree technology. The method is established taking into account the characteristic of flood control and hydropower operation, which is different from the traditional ones, such as the cumulative occurrence frequency method. The results show that the proposed method performs relative well and higher hydropower production are also obtained.(2) A simple rainfall-runoff hydrological model using the 10-day accumulated QPFs from Global Forecast System (GFS-QPFs) Forecasts is presented. Taking the Ertan reservoir as an example, 10-day accumulated GFS-QPFs over Yalong river basin are verified by using a three-category contingency table firstly. Then this paper presents the results from a proposed hydrological model. Results show that inflow forecast errors can be reduced considerably, compared with those from the currently used ARMA model by both quantitative and qualitative analysis. Finally, simulations of medium-range hydropower operation are also presented and simulations demonstrate that, the use of GFS-QPFs has improved in reservoir inflow predictions and hydropower operation of the Ertan hydropower station in the Yalong river basin during the wet season.(3) A new inflow description model is proposed by addressing the uncertainty of inflow and its forecast. The randomness of the inflow is addressed through a posterior transition probability, and the uncertainty in flow forecasts is addressed through both the posterior flow transition probability and the predictive probability of forecasts. Taking the hydropower stations of Yalong river basin as examples, this study details the proposed model including the selection and Discretisation of hydrologic state variables, calculation of conditional probabilities,e.g. the prior and posterior transition probabilities, and the predictive probability of forecast, which will be the important inputs into the optimization operation models.(4) A stochastic optimization operation model is presented to incorporate inflow forecasts with various lead-times as hydrologic state variables. Taking Ertan and Jinping I-Ertan cascade reservoirs for examples, this study details the models including the objective function, the development recursive equations, the steady state polices and theirs simulations. The results demonstrate that including inflow forecasts with various lead-times is beneficial to the Ertan hydropower generation, and the chosen operating policy can not only yield higher hydropower production, but also produce reasonable storage hydrographs effectively.(5) Previous literatures focus on the overall structure and key technologies of reservoirs flood forecast and operation system (RFFOS), researches on system refactoring, however, are rarely done. Refactoring technique is introduced to improve the original RFFOS and a general RFFOS application framework is also given. Real application shows it is much easier for programmer to develop a RFFOS based on these extracted components, which will help the researcher of hydrological engineering to concentrate on the researches on models of flood forecasting and operation optimization policies.Finally, a summary is given and some problems to be further studied are discussed.
|
PDF Full Text Request