To address operational instability issues at hydropower stations, such as frequent unit start-stops and passage through vibration zones, under fluctuating grid loads, a real-time load adjustment strategy was developed to handle grid load fluctuations of varying magnitudes. This strategy flexibly selects the combination of units participating in the adjustment through four modules: power station status updating, maintaining the number of operating units, increasing or decreasing the number of operating units, and global traversal. A load optimization allocation model aimed at minimizing water consumption was established and solved using a dynamic programming algorithm. Taking the Dongfeng Hydropower Station in the Wujiang River Basin as an example, the proposed method was validated. The results indicate that, following optimization, the cumulative number of unit start-stop cycles over three days and the number of instances of crossing the vibration zone decreased by 44.12% and 67.21%, respectively, while the total water consumption decreased by approximately 7.6×105 m³. This real-time load adjustment strategy can effectively ensure power generation demand, significantly improve the economic efficiency and stability of hydropower station operations, and thus provide reliable technical support for the real-time dispatch of hydropower stations under frequent grid load fluctuations.