Process control, PID and adaptive control - Intelligent Buildings and Building Automation - page 133

7.8.1 Auto-tuning
Some adaptive schemes require a priori information about the process
dynamics. For users it would be ideal to have an auto-tuning function in
which the controller can be tuned simply by pushing a button. Auto-tuning
techniques are developed for this purpose. Several manufacturers were
forced to introduce a pre-tune mode to help in obtaining the required prior
information. Auto-tuning of PID controllers is a typical industrial applica-
tion of auto-tuning. The importance of a priori information also appeared
in connection with attempts to develop techniques for automatic tuning of
simple PID regulators. Such regulators, which are standard control stations
for industrial and building automation, are used to control systems with a
wide range of time constants.
Although conventional adaptive schemes seemed to be ideal tools to
provide automatic tuning, they were found to be inadequate because they
required prior knowledge of time scales. Special techniques for automatic
tuning of simple regulators were therefore developed. These techniques are
also useful for providing pre-tuning of more complicated adaptive systems.
An overview of industrial PID controllers with auto-tuning is given below.
Several ways to carry out auto-tuning have been proposed. The most
common method is to conduct a simple test on the process. The test can be
done in open loop or closed loop as described in Section 7.6. In the open-
loop tests the input of the process is excited by a pulse or a couple of steps.
A simple process model, for instance of second order, is then estimated by
using recursive least squares or some other recursive estimation method. If a
second-order process model is estimated, the PID controller can be used to
make pole placement.
The speed and the damping of the system are then the design parameters. A
popular design method is to choose the controller zeros such that they cancel
the two process poles. This gives good responses to set-point changes, while
the response to load disturbances is determined by the open-loop dynam-
ics. The transient response method for automatic tuning of PID controllers
is used in products in the market. It is used also for pre-tuning in adaptive
controllers in some products.
The tuning experiments can also be done in closed loop. A typical example
of this is the self-oscillating method of Ziegler and Nichols or its variants.
The relay auto-tuner based on self-oscillation is used in some products. In
these regulators tuning is initiated simply by pushing the tuning button. One
advantage of making tests in a closed loop is that the output of the process
may be kept within reasonable bounds, which can be difficult for processes
with integrators if the experiment is done in an open loop.
The auto-tuning function is often a built-in feature in standard standalone
PID controllers. Automatic tuning can also be done by using external
equipment. The tuner is then connected to the process and performs a test,
usually in open loop. The tuner then suggests parameter settings, which are