Adaptive Parameter Estimation-Based Predictive Multi-Model Switching Control of Drainage Systems

A discrete-time model with delay was derived, aiming at the sewage overflow and operational cost caused by unexpected time variation and delay in flows of drainage system. Based on the model, an adaptive multi-model switching control strategy using predictive functional control (PFC) is developed. The strategy can estimate the flow to the pumping station and predict the depth of pump well. Combining on-line identification using improved projection algorithm with multi-model switching abates model mismatch and system chattering under the influence of switching over frequently. Initial value matching up to each sub-model was employed to accelerate convergence of estimates, which was identified by off-line estimation. Simulation results indicate that the proposed control approach is robust and effective. The discharge of the downstream pumps is varied automatically by tracking the outflow mainly from upstream pumping stations that the predetermined water level of pump well is maintained by the adoption of this control strategy, which minimizes the overflow pollution.