| Summary: | 碩士 === 國立中正大學 === 化學工程研究所 === 86 === Several chemical processes are of distributed type. Distributed systems are
mathematically described by their partial differential equations. For a linear
distributed system, the Laplace transform of the associated
partial differential
equation is an irrational transfer function containing
transcendental terms. Since an
irrational transfer function has an infinite number of poles, the
dynamic simulation of distributed systems is a very difficult
task. The stability
test and design
of the feedback system for distributed systems are particularly difficult.
The purpose of this thesis is to design optimal PID controllers for linear
distributed systems such that the integral of the square error is minimized.
In fact, this thesis solves an optimal parameter selection problem. In order
to ensure that the selected controller parameters make the overall feedback
control system stable, the D-partition technique is adopted along with the
Nyquist stability criterion to construct the stable domain in the parameter
space.
After the stability domain in the parameter space has been
identified, a crude random
search is performed to choose five sets of PID controller
parameters which are then
served as the initial guesses for the further gradient-based searches.
To illustrate the proposed optimial PID controller design approach, two
numerical examples about the feedback control of a heat transfer process
are worked out.
|
|---|
