Copyright (i) IFAC Digital Control: Past, Present and
Future of PlO Control, Terrassa, Spain, 2000
CONTROL DESIGN FOR PID CONTROLLERS AUTO-TUNING BASED ON IMPROVED
IDENTIFICATION
R. R. Pecharroman and F. L. Pagola
Department oj Electronics and Automatic Control, E. T.S. de fngenieria (lCAf).
Universidad Pontificia Comillas de Madrid, Alberto Aguilera 23, 28015 Madrid, Spain.
Phone 34915422800, ramon@dea.icai.upco.es,pagola@dea.icai.upco.es
Abstract: Every auto-tuning method for PID controllers has two steps: identification of
the process and design of the controller. An identification procedure to obtain two points
of the frequency response of the system was presented in (PecharrOlmm and Pagola,
1999). This paper develops a design method based on the ITAE index. This design
procedure is then related to the results of the process identification, yielding a complete
auto-tuning method that gives very good results in theory and in practice. A complete
test batch is used. The main contribution is that the method is valid for integrating as
well as non-integrating processes. Copyright © 2000 fFAC
Keywords: Autotuners, PID Control, Integral performance indices, Least-squares
approximation, Process models.
I. INTRODUCTION
Most controllers used in industry are of PID type.
Very often, their performance is poor due to, among
other factors, inadequate tuning of the controller
parameters (Astrom and Hiigglund, 1995). Hence,
auto-tuning is a very desirable feature and almost
every industrial PlO controller provides it nowadays.
There are many different auto-tuning methods,
several of them compared in (Hang and Sin, 1991).
An auto-tuning method consists of a process
identification procedure plus a controller design
procedure. Very often, these two aspects are not
clearly differenciated. Most auto-tuning methods fail
in certain cases, but it is not clearly analysed wether
the problem is due to the identification or to the
controller design step. It might be possible to build an
auto-tuning method using the identification procedure
proposed by one author and the controller design
procedure proposed by another one. This paper
85
focuses on the controller design step and on the
integration of both to give a complete auto-tuning
method. The identification step was presented in
(Pecharroman and Pagola, 1999).
A modified PlO algorithm is used, incorporating
valuable advances presented in the literature: two-
degree-of-freedom structure, setpoint weighting, etc.
The controller-parameter tuning is based on the ITAE
performance index, subject to a damping measure, the
maximun sensitivity junction. The optimization is
performed in two steps, allowing simultaneous quasi-
optimization of responses to both setpoint and load
disturbance changes. This method is applied to all the
plants included in the test batch shown in the
Appendix, that intends to represent the industrial
practice. The results are very good.
The next goal is to relate this controller design
method to the results of the identification step: two
points of the frequency response of the process (the