A Novel Model-Free Output-Feedback H_∞ Parameterization Control Method With Unknown States Under Ill-Condition.

Area of Science:

• Control Theory
• Adaptive Dynamic Programming
• Optimal Control

Background:

• Developing model-free H-infinity control is challenging due to unknown system parameters and unmeasurable states.
• Existing methods often require complete system models, limiting their applicability.
• Uncertain disturbances further complicate the design of robust control systems.

Purpose of the Study:

• To propose a model-free output-feedback (OPFB) suboptimal control scheme using adaptive dynamic programming (ADP).
• To achieve H-infinity control under uncertain disturbances without prior knowledge of system dynamics.
• To address challenges posed by unmeasurable states in control system design.

Main Methods:

• Introduced a free matrix to compute suboptimal gain.
• Developed a policy iterative algorithm for solving the algebraic Riccati equation.
• Proposed a model-free ADP algorithm for online learning of control parameters.
• Utilized the Lanczos method to address ill-conditioning in the model-free algorithm.
• Extended the algorithm for systems with unmeasurable states via parameterized reconstruction using input-output data.

Main Results:

• The policy iterative algorithm converges to a solution of the algebraic Riccati equation.
• The model-free ADP algorithm successfully learns control parameters online without system dynamics.
• The extended algorithm effectively handles unmeasurable states using input-output data.
• Simulations on an F-16 aircraft model validated the proposed control scheme's effectiveness.

Conclusions:

• The proposed ADP-based output-feedback control scheme enables model-free H-infinity control.
• The method is effective even with unknown system parameters and unmeasurable states.
• The approach provides a robust solution for control problems under uncertain disturbances.