- 53 views
- 75 downloads
Predictive Control, Estimation and Sensor Placement of Large-Scale Transport-Reaction Systems
-
- Author / Creator
- Zhang, Lu
-
Advanced process control and monitoring rely on the accurate description of complex processes and their dynamic behaviors. Typically, numerous industrial processes are characterized by either partial differential equations (PDEs) or ordinary differential equations (ODEs), depending on whether their dynamics evolve spatiotemporally or temporally. Consequently, these processes are classified as distributed parameter systems (DPS) or lumped parameter systems (LPS). The underlying fact is that most of the large-scale chemical and petrochemical processes are influenced by both temporal and spatial factors, and the existence of the spatial variable in the mathematical model rises challenges in the controller and estimator designs. The focus of this thesis is to develop advanced controllers and observers for the improvement of control performance and reliable monitoring realizations for large-scale DPS processes.
Continuous-time DPS can be considered an important representation of complex industrial processes, while it is more valuable to obtain discrete-time models for the design of controllers and observers when it comes to practical implementation in digital devices. This thesis provides a discrete-time infinite-dimensional modelling framework specifically designed for large-scale DPS while preserving essential model properties (such as stability, observability, input-output mapping and etc.), where no model spatial discretization or spatial model reduction is required.
Considering the inherent complexity and numerous constraints involved in chemical and petrochemical industry processes, the utilization of model predictive control (MPC) offers significant advantages. This thesis introduces an MPC design based on developed discrete-time infinite-dimensional models, aiming to achieve satisfactory performance while handling input and output constraints and addressing the constrained stabilization of disturbed DPS. Additionally, a tracking MPC scheme is formulated for a class of large-scale DPS in a late lumping manner, with the objective of achieving the desired target set-points while accommodating system constraints. Furthermore, to address scenarios where an accurate model of the underlying dynamical system is unavailable, a robust MPC scheme is proposed, incorporating multiple DPS models to ensure system stabilization.
Due to the difficulties in obtaining state information in DPS, state estimation techniques are employed to incorporate controller design. To estimate the spatiotemporal state, the discrete-time Luenberger observer and Kalman filter are proposed for the considered large-scale DPS. To account for the constrained actuator and parameter in the estimation, moving horizon estimation (MHE) is developed by extending the MHE theory of LPS. Additionally, the distributed nature of DPS introduces complexity in terms of sensor placement. This thesis explores the sensor location selection problem along with estimator design accounting for the delayed measurements by minimizing the variance of estimation error.
The effectiveness of the developed discrete-time controllers and estimators are demonstrated by numerical simulations of various large-scale DPS, including tubular reactor, pipeline system, and continuous pulp digester.
-
- Graduation date
- Fall 2023
-
- Type of Item
- Thesis
-
- Degree
- Doctor of Philosophy
-
- License
- This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.