One-shot data-driven controller certification

Area of Science:

• Control Systems Engineering
• Systems and Control Theory

Background:

• Controller certification is crucial for system safety and reliability.
• Existing data-driven methods often require extensive datasets and multiple experiments.
• Theoretical frameworks link controller certification to H∞-norm estimation of specific transfer matrices.

Purpose of the Study:

• To develop a one-shot, purely data-driven method for controller certification.
• To estimate the H∞-norm directly from input-output data using Markov parameters.
• To enable accurate performance estimation with minimal data acquisition.

Main Methods:

• A novel procedure for estimating the H∞-norm directly from closed-loop input-output data.
• Utilizes Markov parameters derived from a single batch of plant data.
• Adaptable for open-loop data acquisition while estimating closed-loop performance.

Main Results:

• The proposed method achieves high accuracy in controller certification using only one data batch.
• Demonstrates superior accuracy compared to existing methods on the same datasets.
• Successfully validated through simulations and experimental results on a three-tank liquid processing plant.

Conclusions:

• The developed method provides an efficient and accurate data-driven approach for controller certification.
• Its ability to use minimal data enhances practical applicability in real-world systems.
• The technique is robust and versatile, applicable to both open-loop and closed-loop scenarios.