Numerical algorithm for a discrete variable gain controller design

Abstract

In this work we present algorithmic design steps of discrete variable gain controllers. The method is derived from the classical discrete deadbeat response approach. It is based on calculating the gain for each discrete sample that achieves the regulation in finite steps number equal the system order. The computation algorithm is implemented numerically by solving a system of nonlinear coupled equations using one of the known evolutionary techniques, in which the genetic algorithm is augmented with Newton-Raphson method. Two industrial control systems are considered for testing the designed controller validity. The efficacy of the method for parameter variations is explored and compared with the conventional approaches.