On LQG self-tuning control: implementation and application

Abstract

The design and construction of a self-tuning controller software package intended for general industrial use as a single-loop controller is presented. An indirect self-tuning regulator is implemented with a Linear Quadratic Gaussian (LQG) control law, and a prediction error parameter estimation algorithm. The control law is derived using a polynomial equation approach. The software design is modular, avoids the use of global variables and includes provision for plant protection. A special feature of the work reported is the emphasis on jacketing software and the user interfaces necessary for routine industrial application. Experimental results demonstrate both the functionality and practicality of the device, and its fail-safe software mechanisms. The field test for the first industrial application is on a Main Boiler Feed Pump (MBFP) speed control sub-loop in a power station application. Some details of this application are given.