REPOSITORY > RESULTS

Abstract

Integrating systems are commonly found in power and paper mills, aerospace control systems, storage tanks, distillation columns, chemical reactors, and petroleum industries. Due to their open-loop instability resulting in an unconstrained output for a constrained input, efficient control of integrating processes is challenging. Control of integrating processes has attracted the interest of researchers for many decades, and several control structures and tuning methods have been developed so far. However, among the existing tuning methods, it is difficult to find one based on a relatively simple experiment and tuning procedure that leads to highly efficient tracking and disturbance rejection control, similar to the magnitude optimum multiple integration (MOMI) tuning method.
In this dissertation, an extension of the MOMI tuning method is proposed for integrating processes controlled by a two-degrees-of-freedom (2-DOF) proportional-integral (PI) and proportional-integral-derivative (PID) controller. The developed method is nonparametric, i.e., it does not require an explicit process model to compute the controller parameters. The controller parameters can be calculated either from measurements in the time-domain or from a process transfer function of arbitrary order with a time delay. Both approaches are exactly equivalent and do not introduce errors in the calculation of the controller parameters.
Furthermore, the developed extension of the tuning procedure provides an additional option to achieve the best overall closed-loop performance (optimal tracking and disturbance-rejection performance) by using the reference weighting parameter or two reference filter structures. Moreover, an additional parameter is provided to modify the speed of the closed-loop response by setting the average closed-loop residence time, i.e., a measure that defines the closed-loop time constant.
Additionally, the stability and robustness of the proposed tuning method to perturbations in the process parameters are studied. The robustness of the measured characteristic areas in a noisy environment is also investigated, and a new approach to reduce the sensitivity of the measured characteristic areas on high-frequency noise is proposed.
In addition to tuning, this thesis develops a method for identifying the lower-order process model either directly from the time-domain measurements or from an arbitrary-order process transfer function with a time delay.
The proposed controllers were also compared with other PI/PID tuning methods for integrating processes. The comparison showed superior control performance compared to the other tuning methods currently proposed. In addition, the proposed tuning method was tested on the following laboratory and industrial setups: a charge amplifier drift compensation system, a laboratory hydraulic system, an industrial autoclave, and a solid oxide fuel cell temperature control system. The closed-loop responses were fast and stable in all experiments.