1-4244-0361-8/06/$20.00 ©2006 IEEE IEEE-TTTC Int Conf on Automation, Quality and Testing, Robotics AQTR, 122-127, (Eds Miclea and Stoian), ISBN 1-4244-0360-X, Cluj-Napoca, Romania, 2006 Development and Evaluation of a PID Auto-Tuning Controller Ioan Naşcu 1 , Robin De Keyser 2 , Silviu Folea 1 , Tudor Buzdugan 1 1 Technical University of Cluj Napoca, Department of Automation, Ioan.Nascu@aut.utcluj.ro , Tudor.Buzdugan@aut.utcluj.ro , Silviu.Folea@aut.utcluj.ro 2 Ghent University, EeSA-department of Electrical Energy, Systems & Automation, rdk@autoctrl.UGent.be Abstract-This paper presents comparative studies regarding a recently developed extension of the widely used relay-feedback PID auto-tuner. The proposed method consists of two steps: process identification and controller design. First, a non-iterative procedure is suggested for identification of two points on the process Nyquist curve. A second-order-plus-dead-time model is obtained and then used for PID controller design based on the internal model principle (IMC). For the identification of the two points on the Nyquist curve a pure relay in the feedback loop (as used in standard auto-tuning) and a relay which operates on the integral of the error are used. The paper compares the performance of autotuning methods with experimental studies on a commercial auto-tuning PID controller - a Shimaden PID auto- tuner. Keywords: auto-tuning, PID controller, relay experiment, Nyquist plot, commercial PID controllers. I. INTRODUCTION At the beginning of the new millennium the PID controller continues to be a key component of industrial control. Their popularity is justified by the following advantages: they have a simple structure, their principle is well understood by instrumentation engineers and their control capabilities have proven to be adequate for most control loops. Moreover, due to process uncertainties, a more sophisticated control scheme is not necessarily more efficient in real-life applications than a well-tuned PID controller. However, it is common that PID controllers are poorly tuned in practice because the choice of controller parameters requires professional – specialized knowledge by the user. To simplify this task and to reduce the tuning-time (process starting up period), PID controllers can incorporate auto-tuning capabilities. The auto-tuners are equipped with a mechanism capable of computing the correct parameters automatically when the regulator is connected to the field [Aström, et al., 1984; Aström, et al., 1992; Aström, et al., 1995; Leva, et al., 2002]. Auto-tuning is a very desirable feature and almost every industrial PID controller provides it nowadays. These features provide easy-to-use controller tuning and have proven to be well accepted among process engineers. For the automatic tuning of the PID controllers, several methods have been proposed. Some of these methods are based on identification of one point of the process frequency response, while the others are based on the knowledge of some characteristic parameters of the open-loop process step response. The identification of a point of the process frequency response can be performed either using a proportional regulator, which brings the closed-loop system to the stability boundary, or by a relay forcing the process output to oscillate. Åström and Hägglund [Åström, et al., 1984] report an important and interesting approach. Their method is based on the Ziegler-Nichols frequency domain design formula. A relay connected in a feedback loop with the process is used in order to determine the critical point. This contribution briefly describes the development of two auto-tuning methods based on the identification of two points on the process Nyquist curve [Nascu and De Keyser, 2003]. The performance of both autotuning methods are compared with those of Åström-Hägglund method and with experimental studies on a commercial auto-tuning PID controller - a Shimaden PID auto-tuner. II. THE EXPERIMENTAL SET-UP A standard closed loop system with single input single output, as shown in figure 1, is considered. To perform relay feedback experiments, the process is first brought to steady-state conditions in manual control or with any preliminary tuned PID controller. The process H p is assumed to be linear, stable and proper. The PID controller has a non-interacting structure cascaded with a first order filter: 1 1 1 1 ) ( + + + s T sT sT K = (s) H f d i c c (1) In this paper it is assumed that the process dynamics can be described with reasonable accuracy by a second-order-plus- dead-time (SOPDT) model as: