Genetic Rules to Tune Proportional + Derivative Controllers for Integrative Processes with Time Delays

摘要

The controller tuning rules proposed by Ziegler & Nichols and their later modifications have been of huge success in the industry due to their simple application and reliability. The acceptance of this technique is based on the proposition of a set of heuristic rules for tuning the controller requiring information obtained from process measurements. The rules involving mathematical models for tuning had not been of the same success because the mathematical model is not available in most cases nor the qualified staff to develop this kind of designs. The intention of the presented article is to follow the approach proposed by Ziegler & Nichols to analyze integrating systems, maintaining the objective of providing a set of tuning rules obtained from the input and output signals. The difference with the previously proposed approach which uses human experience to develop the set of rules, is the use of evolutionary algorithms for synthesizing the tuning rules.