output feedback control of distributed parameter system via adaptive proper orthogonal decomposition

Open Access
Yang, Manda
Graduate Program:
Chemical Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
August 13, 2015
Committee Members:
  • Antonios Armaou, Thesis Advisor
  • distributed parameter system
  • adaptive proper orthogonal decomposition
The control of distributed parameter system (DPS) is of significant importance since many chemical industry processes exhibit spatial variation. In this work, output feedback control is employed to control DPS. Controller design for DPS is a challenge due to the infinite dimensional property of DPS. This issue is circumvented by constructing reduced order models (ROMs). The ROM is generated by Galerkin method, which projects the system into lower dimension space using basis functions. Discrete adaptive proper orthogonal decomposition (DAPOD) is proposed to update the basis functions. A methodology is developed in which DAPOD is coupled with discrete empirical interpolation method (DEIM) to reduce computational burden. We illustrate the performance of this methodology in a diffusion reaction process in which a first order reaction takes place along a rod. The second case is a tubular reactor with recycle in which the objective is to eliminate hot spot. Compared with adaptive proper orthogonal decomposition (APOD), DAPOD provides a more accurate estimation of eigenvalues of covariance matrix and addresses the drawback of modified APOD that new data keep being eliminated.