| Hi-index | 0.00 |
In industrial practice there is an ongoing search for methods that make controller parameter setting easier. Optimization procedures that do not require any mathematical theory are strongly preferred. The most popular existing methods are experimentally based. The classical Ziegler and Nichols method retains its popularity, but some knowledge of tuning is necessary in order to obtain good results in certain cases. The Relay Feedback method, which avoids critical controller setting by making temporary use of a relay in the control loop, has become popular nowadays. However, it is a pity to give up all the available theoretical methods, especially those based on linear control theory, simply due to reluctance to use mathematical modelling in industrial practice. Global optimization methods are an example of theoretically based controller setting that can be used with low requirements on the operator's knowledge. They can be implemented fully automatically without the need for human participation. Although that they were developed for linear models of control loops, they can provide satisfactory results when applied to real control loops. The optimal controller setting indicators that these methods offer, can be evaluated experimentally by additionally exciting the control loop by a sinusoidal signal whose amplitude does not greatly disrupt the controlled process but enables these experiments to be carried out on-line. Some of these evaluation methods are presented here with the motivation to overcome one of the possible disadvantages of the Relay Feedback method - a restricted control function during the identification phase, when critical oscillation are generated in the control loop. The idea underlying the investigation carried out here is to design an adaptive PID controller ready for operating in real conditions and making use of global optimal setting methods.