Abstract: The optimal scheduling of cascade reservoir groups involves joint modeling, efficient solution, rule extraction, benefit evaluation and other aspects. With the joint optimal scheduling of cascade power stations on the Yalong River and the middle and downstream of Jinsha River as the main goal, the large-scale system decomposition coordination and discrete differential dynamic programming method are introduced for optimal solution on the basis of joint scheduling modeling, and a variety of schemes are designed for benefit analysis. Results reveal that the power generation by joint optimal scheduling of cascades is much larger than those by separate scheduling, up by 2.8% on average. Especially in high flow year, joint optimal scheduling has a significant effect on the growth of power generation. The increase in power generation in downstream Jinsha River shares more than 70% of the total increase on average. In low flow year, the increase in power generation of downstream Jinsha River is basically equal to the total increase of the three basins, reflecting the great advantage of cascade reservoirs in downstream Jinsha River in terms of large storage capacity and high water head. In the mean time, the joint optimal power generation of the middle and downstream Jinsha River is larger than that of Yalong River and downstream Jinsha River. It would be of great significance to explore the joint optimal scheduling of cascade power station groups in the middle and downstream of Jinsha River through investment mergers and acquisitions, dispatching compensation and other forms.

Key words: cascades on Yalong River, cascades on middle and lower Jinsha River, joint optimal scheduling, large-scale system decomposition coordination, discrete differential dynamic programming methods