Research On Model Predictive Control Of Urban Drainage System

Urban drainage system collects a combination of wastewater and rainfall runoff. The sewage should be transported to the treatments by sewers, but overflows are sometimes caused during rainstorm events due to some restrictions of the system, such as storage capacity of sewers, drainage capacity of pumps, receiving capacity of treatments and so on. Real-time control of urban drainage system can effectively improve the system performance, reduce the urban drainage system overflow and control energy consumption. This thesis studies model predictive control of the urban drainage system, including overflow control and water-level control. The main purpose of overflow control is to reduce the system overflow during heavy-intensity rainfalls. Water-level control reduces the control energy consumption by maintaining the water-level at a set-point.The thesis introduces the mathematical model of the urban drainage system and its simplified formulation. And then based on it, different model predictive control algorithms for overflow control and water-level control are developed respectively. As for overflow model, in order to reduce on-line computation for large-scale urban drainage systems, this paper proposes two strategies: aggregation strategy and decomposition-coordination method. As for water-level model, an off-line synthesis algorithm of robust model predictive control based on polyhedron invariant set is proposed. This method not only enlarges the initial feasible region but also improves the control performance significantly. Finally the simulation results illustrate the advantages of the proposed methods.