Reduced-Order ESO-Based Predictor Control for Time-Delay Processes

Abstract

This paper proposes a control structure for SISO any-order time-delay processes. The proposed approach combines predictor-based strategies, known for its enhanced state and disturbance prediction capabilities, with a reduced-order extended state observer (ESO). To enhance robustness, a direct relationship is established between the proposed predictor-based controller and robust dead-time compensators (DTCs), while preserving the state and disturbance estimation capabilities of observer-based predictors. This relationship allows the use of well-established DTC principles for robustness analysis and controller tuning. The controller design is divided into three main parts. First, the reduced-order ESO is designed to estimate the unmeasurable states and disturbances of the process. Then, the predictor is proposed to estimate the future values of the process states and disturbances. Finally, a controller is designed for the delay-free process model, using either a PI/PID structure or a state-feedback law based on the predicted process state. The tuning procedure is derived based on the desired setpoint-tracking performance and a trade-off between disturbance attenuation and closed-loop robustness, using frequency-response techniques. The proposed controller can be applied to open-loop stable, unstable, and integrating processes. Finally, simulations compare the method with recent controllers from the literature, and an experimental study on the temperature control of a neonatal incubator prototype demonstrates its practical applicability.