Robust fractional-order auto-tuning for highly-coupled MIMO systems.

Area of Science:

• Control Engineering
• Industrial Automation
• System Dynamics

Background:

• Industrial processes often involve complex, highly-coupled Multiple-Input Multiple-Output (MIMO) systems.
• Tuning controllers for these systems requires methods that ensure robustness against model uncertainties and process variations.

Purpose of the Study:

• To propose a methodology for tuning fractional-order PI controllers for MIMO systems using the Kissing Circle (KC) tuning method.
• To compare the performance and robustness of a fractional-order PI controller with an integer-order PI controller for a real-world industrial process.

Main Methods:

• The Kissing Circle (KC) tuning method, a model-free approach relying on frequency domain specifications, was employed.
• A direct low-order discretization method was used for implementing the controllers on an experimental landscape office lighting system.
• Both controllers were evaluated through simulations and experimental tests, including exposure to process variations.

Main Results:

• The fractional-order PI controller demonstrated superior robustness, managing an 85% time constant mismatch compared to 79% for the integer-order controller.
• Experimental results showed similar control performance between the fractional-order and integer-order PI controllers based on Integer Absolute Error.
• A 6% reduction in power consumption was observed with the fractional-order controller.

Conclusions:

• The proposed KC-based methodology effectively tunes fractional-order PI controllers for MIMO systems, enhancing robustness without sacrificing control performance.
• The auto-tuning nature of the KC method simplifies controller implementation for industrial applications.
• The fractional-order controller offers improved robustness and energy efficiency, crucial for managing uncertainties in MIMO systems.