Data-Driven Dynamic Internal Model Control.

Area of Science:

• Control Engineering
• System Identification
• Data-Driven Control

Background:

• Traditional model-based internal model control (IMC) requires explicit system models, which are difficult to obtain for complex nonlinear nonaffine systems.
• Existing data-driven methods often struggle with nonlinearities and nonaffine structures.

Purpose of the Study:

• To propose a novel data-driven dynamic internal model control (D3IMC) scheme for unknown nonlinear nonaffine systems.
• To bypass the need for explicit system modeling in control design.
• To enhance robustness against model uncertainties and external disturbances.

Main Methods:

• Development of a dynamic internal model (DIM) using input-output data via a compact dynamic linearization approach.
• Proposal of D3IMC with nominal and uncertainty compensation algorithms.
• Incorporation of an adaptive parameter updating law for robustness.
• Extension to a full-form dynamic linearization-based D3IMC for complex dynamics.

Main Results:

• The D3IMC scheme effectively controls unknown nonlinear nonaffine systems without explicit modeling.
• The nominal control algorithm ensures rapid response to feedback errors.
• The uncertainty compensation algorithm effectively addresses model-plant mismatch and external disturbances.
• Adaptive mechanisms provide inherent robustness against system uncertainties.

Conclusions:

• The proposed D3IMC methods are significant advancements over traditional model-based IMC.
• These data-driven approaches offer a viable alternative for controlling complex systems where explicit modeling is challenging.
• Simulation studies validate the effectiveness and robustness of the developed D3IMC schemes.