In this paper, the Bayesian Forecasting System ( BFS) is presented to address the uncertainties of inflow forecast and precipitation forecast as well as to reduce the complexity of the reservoir hydropower generation model. The BFS, based on Bayesian decision theory and total probability formula, is applied to quantitatively depict the uncertainties of inflow forecast by analyzing the uncertainties of hydrological model, inflow and precipitation forecast. Furthermore, the BFS is also used to develop a deterministic inflow forecast ( DIF ) , the probabilistic quantitative inflow forecast ( PQIF ) and expectations forecast. Besides that, Stochastic Dynamic Programming ( SDP ) and Bayesian Stochastic Dynamic Programming ( BSDP) are combined to formulate the hydroelectric operating policies. Finally, by using information of the 10 days lead time precipitation forecast of Global Forecast System ( GFS) and taking Huanren Reservoir as a study case, the simulation of hydropower generation operating for this reservoir is provided. The simulation results show that the power generation dispatching based on the combination of BFS with SDP model can effectively improve the power generation efficiency and stability of the reservoir.%为了充分考虑降雨、径流预报的不确定性和降低水库发电调度模型的复杂性，采用贝叶斯概率水文预报系统( BFS)耦合降雨预报的不确定性和径流预报模型本身的不确定性来定量描述径流预报的不确定性，发布径流确定性预报、概率预报和概率预报期望值；结合随机动态规划( SDP)模型和贝叶斯随机动态规划( BSDP)模型来制定发电调度图；以浑江桓仁水库流域为背景，采用美国国家天气局的全球预报系统( GFS)发布的10 d降雨预报信息作为预报模型输入，模拟桓仁水库的发电调度过程。模拟结果表明基于径流贝叶斯概率预报的水库发电调度能有效提高水库的发电效益和保证率。
展开▼
