Abstract: For the Eastern Route of the South-to-North Water Diversion Project, the prediction accuracy of its medium and long-term water inflows from the major water sources is not high enough, and the forecast period is limited. Under this circumstance, how to conduct scientific scheduling and rolling decision-making for such a complex systems is the key to a full realization of the comprehensive benefits of the Jiangsu section of this route. This paper presents a new two-stage stochastic optimization scheduling method based on a residual period scheduling cost function, which can realize a medium and long-term rolling scheduling decision for the project under the condition of forecast uncertainty. First, we examine scheduling operation rules for each project unit, and work out a calculation method of system water transfer cost by the principle of water balance. Then, we reveal the trend of historical water inflows from the main water sources, and derive a residual period cost function as a quantitative description of the residual period water transfer cost. Finally, a two-stage decision-making model is constructed for different inflows, and a medium and long-term flow scheduling scheme is generated in a rolling manner. The results show that this two-stage model can overcome the adverse effects of medium and long-term forecast uncertainty on dispatching decision-making, and reduce the cost of water transfer significantly through adjusting mutual across multiple water sources and selecting water sources and transfer routes.

Key words: uncertainty, cross-basin water transfer, two-stage decision-making, medium and long-term scheduling, residual cost