HD-MPC: hierarchical distributed model predictive control for the great lakes flow network optimization

With the rapid development of smart city construction, smart water resource management, as an important part of urban infrastructure optimization, is in dire need of efficient and adaptive control techniques to meet the challenges of complex dynamic systems. In this study, a hierarchical distributed model predictive control (HD-MPC) system based on Alternating Direction Multiplier Method (ADMM) coordination is proposed for the Great Lakes, an important water network spanning the U.S.-Canada border, in conjunction with the need for multi-source data fusion and distributed intelligent decision making in smart cities. The system solves the multiple-input multiple-output (MIMO) water level regulation problem in the Great Lakes flow network by integrating multi-objective optimization and dynamic feedback mechanisms. By integrating hydrologic, climate, and stakeholder demand data, an optimization model based on genetic algorithm and hierarchical analysis (GA-AHP) was constructed, and ADMM was used to achieve distributed collaborative control. Simulations based on 2012-2022 historical data show that HD-MPC can effectively stabilize water level and adapt to extreme climate, providing theoretical support and technical reference for the dynamic optimization of large-scale water resource systems in smart cities.