Hierarchical predictive control of integrated wastewater treatment systems

摘要

The paper proposes an approach to designing the control structure and algorithms for optimising control of integrated wastewater treatment plant-sewer systems (I WWTS) under a full range of disturbance inputs. The optimised control of I WWTS allows for significant cost savings, fulfilling the effluent discharge limits over a long period and maintaining the system in sustainable operation. Due to the specific features of a wastewater system a hierarchical control structure is applied. The functional decomposition leads to three control layers: supervisory, optimising and follow-up. A temporal decomposition that is applied in order to efficiently accommodate the system's multiple time scales leads to further decomposition of the optimising control layer into three control sublayers: slow, medium, and fast. An extended Kalman Filter is used to carry out an estimation of needed but not measured plant states in real time. The robustly feasible model predictive controller produces manipulated variable trajectories based on a dedicated grey box (GB) model of the biological processes and drawing its physical reality from the well known ASM2d model. The GB model parameters are dependant on the plant operating point and therefore are continuously estimated. As it is impossible to efficiently control the plant under all influent conditions that may occur by using one universal control strategy, different control strategies are designed. Recently developed mechanisms for soft switching between the MPC control strategies are applied in order to smooth the state and control transient processes during the switching. The methodologies and algorithms proposed in the paper are validated by simulation based on real data records from a wastewater system located in Kartuzy, northern Poland. The control system was implemented at the case-study site to generate in real time the control actions that were assessed by the plant operators and verified by simulation based on a calibrated plant model.