Are heuristic methods for PID controller design reliable?
The main purpose of this paper is to define the set of admissible systems and test the existing methods on some extreme systems of this set. It is shown, that each method fails for some chosen admissible systems.
Over the years, there are many simple empirical or semi-empirical methods developed for PID tuning based mainly on the step response knowledge. Usually, the tuning formulas are not theoretically based, so we can not define the sphere of usage. These methods are often failing in practise, because of the difference between the real system and the testing system used for derivation of the formulas (usually first order plus deadtime system). The popular kt method has been tested on the set of processes, which looks very large, but does not cover all possible real processes. The main purpose of this paper is to define the set of admissible systems and test the existing methods on some extreme systems of this set. It is shown, that each method fails for some chosen admissible systems.
Download
methodrevision.pdf (364 Kb)
MethodRevisionPosterA0.pdf (1202 Kb)
RSS