Data-driven PID Gain Tuning from Regulatory Control Data Based on Generalized Minimum Variance Evaluation

Abstract

This paper considers a PID gain tuning method based on generalized minimum variance evaluation. The method derives the PID gains for reducing the variance of the generalized output from regulatory control data disturbed by colored noise. Advantages of the method include the use of normal operating data with no additional experiment for PID tuning. Numerical examples for a CARIMA (Controlled Auto-Regressive Integrated Moving Average) model and a liquid level control model show the proposed method is effective for not only colored noise but also stochastic disturbances combined with periodical rectangular deterministic signals. The simulation results demonstrate that the proposed approach has robustness to model mismatch of disturbance characteristics.