Disturbance-rejection-based tuning of proportional-integral-derivative controllers by exploiting closed-loop plant...

Area of Science:

• Control Engineering
• Process Systems Engineering
• Automation and Control Theory

Background:

• Traditional PID controller tuning often relies on accurate process models, which can be difficult to obtain or may not represent the system accurately under varying conditions.
• Disturbance attenuation is a critical performance metric in many industrial control systems, directly impacting product quality and operational efficiency.
• Existing methods may lack robustness to measurement noise or require online retuning for optimal performance across diverse plant dynamics.

Purpose of the Study:

• To propose a systematic, data-based design method for tuning proportional-integral-derivative (PID) controllers specifically for disturbance attenuation.
• To develop algorithms that directly utilize closed-loop plant data, eliminating the need for a prior process model.
• To enable robust and flexible PID controller tuning applicable to a wide range of stable, integrating, and unstable plants, including online retuning.

Main Methods:

• A data-driven approach is employed, exploiting closed-loop plant data without requiring a process model.
• Two algorithms are developed: one optimizes the reference model's time delay via a nonlinear optimization problem, and the other uses an approximation for analytical PID tuning formulas.
• The methods incorporate plant data integrals into regression equations, ensuring robustness against measurement noise and including an adjustable parameter for performance-robustness trade-offs.

Main Results:

• The proposed method successfully derives PID controller parameters that minimize deviations from a reference model for disturbance rejection.
• Both developed algorithms demonstrated robustness to measurement noise due to the use of plant data integrals.
• Simulation examples across various process dynamics, including reactor systems, validated the effectiveness of the proposed tuning method.

Conclusions:

• The proposed data-based PID tuning method offers a model-free, robust, and flexible approach to disturbance attenuation.
• The ability to tune controllers online makes the method suitable for improving existing underperforming control systems across different plant types.
• The adjustable design parameter provides users with control over the balance between performance and robustness.