The regulation of large-scale cascade hydropower stations in dry season are confronted with complex market environment and changeable inflow situations during the dry-wet transition period. To address this issue, a multi-mode adaptive matching method for the operation decision of cascade hydropower stations is established. Under “no (drop) pressure environment” mode and “with (drop) pressure environment” mode, we constructed an optimal scheduling model, which can select and switch scheduling mode and target automatically and flexibly according to the forecast information, and generate dynamic control and flexible adjustment strategies for the water level at each hydropower station. The simulation results show that under the mode of “no (drop) pressure environment”, the power generation increased by 2.57%; under the mode of “with (drop) pressure environment”, the power generation increased by 4.59% and 5.32% under the approaches of “uniform breakthrough” and “centralized breakthrough”, respectively. The former mode faces less risk of failure to reduce water level in place, while the latter generates greater power benefits. The model can effectively reduce the risk of water level drop and power generation in dry season, as well as improve the overall power generation benefit of the basin. It has excellent adaptability and optimization effect for different inflows, power generation and other working conditions.