Artificial intelligence based gain scheduling of PI speed controller in DC motor drives

Abstract

The paper analyses applicability of different artificial intelligence based gain scheduling techniques for a conventional PI controller. Three different methods are elaborated. These are the artificial neural network based gain scheduling, gain scheduling by means of an adaptive neuro-fuzzy inference system, and gain scheduling using a self-constructing Takagi-Sugeno fuzzy rule-based system. All the three methods are applied to gain scheduling of a PI speed controller in a DC motor drive. A comparative analysis of the drive performance with PI speed controller without gain scheduling and with PI speed controller with gain scheduling, using the three described gain schedulers, is performed. Good quality of performance is achieved over a wide range of operating conditions with all the three methods of gain scheduling.