Logo image
Nonlinear model predictive control of processes with incomplete state measurements
Thesis   Open access

Nonlinear model predictive control of processes with incomplete state measurements

Mohammed Alhajeri
Master of Science (M.S.), Drexel University
Aug 2018
DOI:
https://doi.org/10.17918/aysb-hs98
pdf
Alhajeri_Mohammed_20181.59 MBDownloadView

Abstract

Self-tuning controllers Predictive control--Mathematical models Chemical Engineering
This dissertation consists of two main parts. The first part provides a comprehensive review of the tuning guidelines that have been introduced for model predictive controllers (MPCs) since 2010. The second part deals with nonlinear control of single-input single-output processes with manipulated-input-saturation constraints, incomplete state measurements, and an unstable steady-state operating point. A nonlinear controller is proposed. It includes an input-output linearizing state feedback controller, which is also an analytical solution to a shortest prediction horizon, continuous-time MPC optimization problem. It uses a closed-loop nonlinear reduced-order state observer to estimate unmeasured state variables. As it handles input constraints optimally, it exhibits no integrator windup. Given the closed-loop stability of the control system, it guarantees zero steady-state error (offset) in the presence of constant process disturbances and process-model mismatch. Its application and performance are illustrated by applying the controller to two nonlinear chemical process examples, a chemical reactor and a bioreactor, via numerical simulations.

Metrics

46 File views/ downloads
33 Record Views

Details

Logo image