A new modified Smith predictor: the concept, design and tuning.

Area of Science:

• Control Systems Engineering
• Process Control
• Automation

Background:

• Industrial processes often exhibit long time delays, posing significant challenges for effective control.
• Existing control strategies like the double controller-scheme (DCS) and predictive PI controllers have limitations in balancing performance and robustness.

Purpose of the Study:

• To introduce a novel modified Smith predictor (MSP) designed for enhanced performance in processes with long time delays.
• To present a tuning methodology for the MSP that optimizes its performance characteristics.
• To evaluate the efficacy of the MSP against established controllers for time-delay systems.

Main Methods:

• The modified Smith predictor (MSP) is developed as an extension of the double controller-scheme (DCS).
• A tuning approach combining the magnitude optimum criterion with process parameterization using multiple integrals of the open-loop step response is employed.
• Performance evaluation is conducted through simulation studies, comparing the MSP with DCS and Hägglund's predictive PI controller.

Main Results:

• The MSP demonstrates superior performance compared to both the DCS and Hägglund's predictive PI controller in simulation studies.
• The MSP allows for adjustable trade-offs between disturbance rejection and robustness to process parameter variations via a single parameter.
• The proposed tuning method effectively optimizes the MSP's performance for time-delay systems.

Conclusions:

• The modified Smith predictor (MSP) presents a significant advancement for controlling processes with long time delays.
• The MSP offers a flexible and effective solution for balancing disturbance rejection and robustness.
• The developed tuning strategy enhances the practical applicability of the MSP in industrial settings.