Multicriterion modelless PID controller autotuning keeping desired values of control quality indicators

Abstract

The development of the autotuning method presented here was motivated by the demands of industrial practice. Industry prefers to use PID controllers and to make experimental settings on-line. Usually used tuning methods have two disadvantages: interruption or restriction of control function while the controller is being set and the use of Ziegler and Nichols rules, which leads to stable but oscillating responses. The paper describes one way of experimentally obtaining the control quality indicators based on frequency response evaluation and using them in a new optimal controller parameter setting computation procedure without control function interruption.