Model predictive control of a multivariable soil heating process

Roy, Prodyut Kumer (2005) Model predictive control of a multivariable soil heating process. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (3MB)

Abstract

Multivariable control has been a challenging research area in process control, particularly for dynamically coupled and nonlinear time varying process systems. Since 1960, various multivariable control techniques have been proposed in the literature to address these issues. Out of these techniques Model Predictive Control (MPC) based control methodologies has received considerable attention during last few decades. -- The aim of this thesis is to provide a comprehensive analysis of different MPC techniques that can be used for a wider class of multivariable process systems. MPC schemes use a model to predict the future behavior of the process to be controlled and the control move that provides the minimum future error is chosen to drive the system. The model employed in the MPC scheme is generally a linear model. The representation of the linear model in two different forms, parametric form or weighting sequence form, has developed two popular and widely accepted MPC techniques, such as Generalized Predictive Control (GPC) and Dynamic Matrix Control (DMC) based MPC techniques. Although the GPC representation is the most advanced form of MPC, the DMC technique is popular in industrial applications. The strict linear representation of the process model in the above MPC schemes is insufficient to provide better response results against nonlinear and time varying systems. To overcome this issue, two approaches are incorporated: (a) adaptive MPC design and (b) fuzzy modeling. The adaptive structure uses an online parameter identification technique using the Recursive Least Squares (RLS) method. The fuzzy MPC system uses the Takagi-Sugeno (TS) type fuzzy rule based model structure. Each rule of the TS system represents a local linear model of the process. This particular feature is exploited to extract the linearaized parameters of the fuzzy model in order to define an adaptive fuzzy MPC system using the RLS technique. The performances of the two adaptive MPC schemes are verified against a simulated multivariable nonlinear soil heating process system. The control objective is to maintain a desired temperature profile of the soil heating system, while tracking the temperatures outputs at three different locations in the soil sample. Three heaters are located at the outer surface of the soil cell and considered as point heat sources in the model. The soil heating system is modeled using the general purpose ABAQUS finite element program and is dynamically linked with the FORTRAN based control code to achieve a realistic simulation. In order to show the effectiveness, the performances of the proposed control schemes are compared against the tracking performances of the linear model-based non-adaptive MPC techniques. A decoupled multivariable PID control scheme is also developed in this study to justify the superiority of the MPC based control strategies. The simulations results suggest the superior performance of the proposed adaptive MPC schemes against other linear MPC techniques.

Actions (login required)

View Item