PID Controllers, 2nd Edition

Chapter 5.1 - New Tuning Methods: Introduction

Many methods for designing PID controllers were presented in Chapter 4. From this we can conclude that there are many issues that have to be taken into account when designing a controller, e.g., load disturbance response, measurement noise, setpoint following, model requirements, and model uncertainty. It is also clear that there is a need for a variety of tuning methods; simple techniques that require little process knowledge as well as more elaborate methods that use more information about the process.

In this chapter we use the insight obtained in Chapter 4 to develop new methods for controller tuning. In Section 5.2 we discuss the key requirements on a good design method. In addition to the issues mentioned above, we consider the choice of design parameters and the process knowledge required. The method used to develop the new rules is quite straightforward. First we apply a reliable design method with the desired characteristics to a large test batch of representative processes. Then we try to correlate the controller parameters obtained with simple features that characterize the process dynamics.

In Sections 5.3 and 5.4 we present tuning rules that can be viewed as extensions of the Ziegler-Nichols rules. The main difference is that we are using three parameters to characterize process dynamics instead of two parameters used by Ziegler and Nichols. It is shown that the new methods give substantial improvements in control performance while retaining much of the simplicity of the Ziegler-Nichols rules.

Methods based on the step response of the process are presented in Section 5.3. In this case we characterize process dynamics with the parameters a and L used by Ziegler and Nichols and, in addition, the normalized dead time t. These parameters are easily determined. The tuning rules obtained give the normal PID parameters and, in addition, the setpoint weighting.

In Section 5.4 we present frequency-domain methods. They are based on the parameters ultimate gain Ku, ultimate period Tu and gain ratio ?. These parameters can be obtained from the conventional Ziegler-Nichols experiment or an experiment with relay tuning combined with a determination of the static gain of the process.

The methods used in Sections 5.3 and 5.4 are based on approximate process models. In Section 5.5 we present an efficient method of computing controller gains when the transfer function of the process in known. These results make it possible to judge the advantage in obtaining more process information.

In Section 5.6 we explore some consequences of the results of the previous sections. The closed-loop systems obtained with the new tuning rules may have many poles and zeros. The behavior of the closed-loop system, however, is dominated by a few poles and zeros. These can be related to the key features of the process. By investigating these relations we get interesting insight into the properties of the closed-loop system, which can be used to judge achievable performance directly from the process features.

Examples of using the new tuning rules are given in Section 5.7.

UNLIMITED FREE
ACCESS
TO THE WORLD'S BEST IDEAS

SUBMIT
Already a GlobalSpec user? Log in.

This is embarrasing...

An error occurred while processing the form. Please try again in a few minutes.

Customize Your GlobalSpec Experience

Category: Motion Controllers
Finish!
Privacy Policy

This is embarrasing...

An error occurred while processing the form. Please try again in a few minutes.