Efficiency improvement in power generation and navigation for grid-connected hydropower stations has emerged as a significant concern due to challenges such as discrepancies between declared and actual ship arrival times, as well as unstable power generation. To address these issues, this paper proposes a multi-objective real-time scheduling model. The proposed model incorporates an energy storage mechanism and prediction mechanism .Using XGBoost, the actual arrival time of the boat to the water station is predicted. Furthermore, an energy storage mechanism is introduced to stabilize power generation by charging the power storage equipment during surplus generation and discharging it during periods of insufficient generation at the hydropower station. We design a new scheduling objective metric. The proposed model employs the Non-Dominated Sorting Beluga Whale Optimization (NSBWO) algorithm, which combines the Elitist Non-Dominated Sorting Genetic Algorithm (NSGA-II) and Beluga Whale Optimization (BWO), to optimize and solve the real-time discharge flow scheduling in hydropower stations across different time periods.Experimental results demonstrate the superiority of the proposed model in ship expected arrival time prediction and downstream flow scheduling, exhibiting higher prediction accuracy and better solutions. This research contributes a theoretical basis and provides a practical reference solution for addressing hydropower generation scheduling problems.