Comparison between GPC and adaptive GPC based on Takagi Sugeno multi-model for an Activated Sludge Reactor

This paper investigates the use of Adaptive Generalized Predictive Control (TS-AGPC) for an activated sludge reactor, based on a Takagi Sugeno (TS) model, and presents the comparison between the latter and Generalized Predictive Control using an overall TS model (TS-GPC). The reduced bio-reactor Activated Sludge ASM1 Model, which describes the biological degradation of an activated sludge reactor, is designed based on several simplifications, as a TS model, its structure being based on a set of linear submodels, covering the process input-output space, interpolated by a nonlinear weighting function µ. The adaptive GPC approach is obtained by switching between linear submodels of the TS formulation. This is performed by selecting, in turns, a portion of the weighting function µ. The winning model will then act as an internal model for the TS-AGPC control law formulation, whereas the complete TS model is used in the calculation of the TS-GPC control law. Finally, the performance under input and parametric disturbances as well as control variable constraints of the TS-AGPC controller are compared to those for a global TS-GPC controller and a benchmark PID in terms of error and response dynamics.

---

Opracowanie rekordu w ramach umowy 509/P-DUN/2018 ze środków MNiSW przeznaczonych na działalność upowszechniającą naukę (2018).