数字孪生技术在大型水库调度中的实时仿真与智能决策支持系统研究

Key Points

• This research aims to explore the use of digital twin technology for real-time simulation and decision-making in reservoir management.
• Developed a real-time mapping mechanism between physical reservoirs and virtual models
• Integrated multi-source sensor data along with high-fidelity hydrological models
• Utilized LSTM for rainfall prediction and GPU parallel computing to improve simulation efficiency and accuracy
• Applied DDPG algorithm for dynamic gate control within a multi-objective optimization framework
• Enhanced decision transparency through 3D visualization and SHAP explanation modules
• Successfully simulated flood evolution and sediment deposition processes
• Achieved improved simulation efficiency and accuracy through advanced computational techniques
• Enabled dynamic control of reservoirs balancing flood safety, water supply benefits, and ecological protection
• Demonstrated enhanced decision-making transparency with novel visualization and explanation tools