In order to analyze the inflow process of large reservoirs in the upper reaches of the Yellow River and the time history variation law of cascade power generation under the influence of climate change, the Longyangxia and Liujiaxia cascade reservoirs in the upper reaches of the Yellow River were taken as examples, and Mann-Kendall method was used to identify the abrupt change of the Tangnaihai and Xiaochuan hydrological series. On this basis, the HBV model considering the snow melting process was constructed, and the statistical downscaling method ( SDSM) was used to analyze the CanESM2 and GFDL _ ESM2G climate models. The rainfall and temperature data of three climate change scenarios (RCP2.6, RCP4. 5 and RCP8. 5) was downscaled in space and was used to drive hydrological model to predict future runoff. Finally, a cascade joint generation scheduling model in the upper Yellow River was constructed to reveal the impact of climate change on future generation scheduling process. The results show that the mutation years of runoff series in the upper Yellow River are concentrated in the 1980s, and the runoff decreases significantly after 2000. From 2021 to 2050, climate change will increase the runoff from June to September, and decrease the runoff in non-flood period. With the passage of time, the change law of cascade power generation of the Longyangxia and Liujiaxia Reservoirs is different. Under RCP8. 5, the uncertainty of climate model has the greatest impact on the difference.