Data-Driven Internal Model Control for Output Regulation

Area of Science:

• Control Theory
• Data-Driven Control
• Systems Engineering

Background:

• Output regulation is a core control theory problem, traditionally requiring known system models.
• Data-driven control offers solutions for unknown systems, but struggles with noisy data in existing methods.
• Existing data-driven output regulator equations (OREs) fail to achieve zero tracking error with noisy data.

Purpose of the Study:

• To address the output regulation problem for unknown single and multiagent systems (MASs) using noisy data.
• To overcome limitations of existing data-driven methods by incorporating the internal model principle.
• To achieve exact output regulation (zero tracking error) in data-driven control frameworks.

Main Methods:

• Leveraging Willems et al.'s fundamental lemma for data-driven control.
• Applying the internal model principle to robust output regulation.
• Solving data-based linear matrix inequalities (LMIs) for controller design.
• Extending the framework to nonlinear systems and multiagent systems (MASs).

Main Results:

• Exact output regulation (zero tracking error) is achieved for linear time-invariant (LTI) systems via data-based LMIs.
• The proposed framework is successfully extended to nonlinear systems and both linear and nonlinear MASs.
• Numerical tests confirm the efficacy of the developed data-driven controllers.

Conclusions:

• The internal model principle, combined with data-based LMIs, provides an effective solution for output regulation with noisy data.
• This data-driven approach enables precise control of unknown systems, including complex multiagent systems.
• The study advances data-driven control by enabling zero tracking error in challenging scenarios.