Comparison of practical adaptive algorithms in pH control

The physical and chemical properties of practical pH processes are usually not exactly known. Sophisticated adaptive pH control methods that are based on explicit process models are vulnerable to model uncertainties and their implementation in practice may be difficult.

This paper studies four strategies in the control of a pilot plant pH process that take the uncertain nature of the process into account. The first one is a standard PID controller, the second is based on an algorithm developed by Kurz, the third is a self-organising fuzzy controller, and the fourth is a multi-model controller that is based on a self-organising map. The results show that all the presented adaptive algorithms based on local instantaneous models outperform the non-adaptive PID. Furthermore, it seems that the multi-model approach is especially well suited for practical pH process control with rapidly varying buffering properties.