A new approach to auto-tuning of PID controllers

Abstract

Delayed state variable filters are implemented for the estimation of parameters of monotone open-loop processes that can be approximated by a first-order lag coupled with a time delay. Based on the delayed filtered step output response, and the first and second-order derivatives, a new estimation procedure has been derived for on-line estimation of the corresponding static process gain, the apparent dead time and the apparent time constant. A new PID tuning formula is formed that is based on the placement of the open-loop zeros related to the apparent time constant and time delay, to give a controlled system with a damping of 0-5 or a maximum overshoot of approximately 16-3%. Extensive simulation studies have also been carried out to relate the controlled output step response to the open-loop system parameters, which could then be re-applied in closed loop retuning of PID controllers. Efforts have also been made to apply the tuning technique to the control of multivariable systems. Simulations are included to illustrate the concepts.