PI/PID Controller Relay Experiment Auto-Tuning with Extended Kalman Filter and Second-Order Generalized Integrator as Parameter Estimators

Abstract

This paper presents a method for the estimation of key parameters of limit cycle oscillations (amplitude and frequency) during a relay experiment used for automatic tuning of proportional-integral (PI) and proportional-integral-derivative (PID) feedback controllers. The limit cycle parameter estimator is based on the first-order extended Kalman filter (EKF) for resonance frequency estimation, to which a second-order generalized integrator (SOGI) is cascaded for the purpose of limit cycle amplitude estimation. Based on thus-obtained parameters of the limit cycle oscillations, the ultimate gain and the ultimate period of the limit cycle oscillations are estimated. These are subsequently used for the tuning of PI and PID controller according to Takahashi modifications of Ziegler-Nichols tuning rules. The proposed PI and PID controller auto-tuning method is verified by means of simulations and experimentally on the heat and air-flow experimental setup for the case of air temperature feedback control. The results have shown that the proposed auto-tuning system based on relay control experiment for the heat and air-flow process PI/PID temperature control can capture the ultimate gain and period parameters fairly quickly in simulations and in experiments. Subsequent controller tuning according to Takahashi modifications of Ziegler-Nichols rules using thus-obtained ultimate point parameters can provide favourable closed-loop load disturbance rejection, particularly in the case of PID controller.