Real-time intelligent regulation of gate-pump groups in open channels of water transfer projects under water diversion disturbances during the icing period

Key Points

• The research aims to develop an intelligent scheduling model for gate-pump groups in open channels, specifically addressing challenges during the icing period.
• Conducted hydraulic simulations of gate-pump responses to step disturbances in water transfer projects.
• Utilized deep reinforcement learning to create a real-time scheduling model capable of managing nonlinear, high-dimensional scenarios.
• Tested the model's robustness under varying operational conditions.
• Achieved a significant elevation in water levels while ensuring operational safety and minimizing gate adjustments.
• Identified safety thresholds for disturbance during the icing period, enhancing control strategies.