Global Prescribed Performance Control for Strict Feedback Systems Pursuing Uncertain Target.

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Neural Networks

Background:

• Real-time trajectory prediction and control are challenging for systems with uncertainties.
• Existing methods often require prior target modeling or offline training.
• Prescribed performance control (PPC) typically needs strict initial condition constraints.

Purpose of the Study:

• To develop an online solution for reconstructing and predicting uncertain target trajectories in real-time.
• To propose an adaptive tracking control scheme for strict-feedback systems with unknown control directions and prescribed performance.
• To overcome limitations of existing methods regarding prior modeling, offline training, and initial condition constraints.

Main Methods:

• Utilized a general regression neural network (GRNN) for online trajectory reconstruction and prediction.
• Developed an adaptive tracking control scheme incorporating novel time-varying scaling and translation functions.
• Ensured global stability without strict initial condition requirements.

Main Results:

• The proposed method enables real-time reconstruction and prediction of uncertain target trajectories.
• The adaptive control scheme guarantees prescribed performance for strict-feedback systems.
• Tracking errors converge to a small, pre-defined set within a specified time, despite uncertainties and unknown control directions.

Conclusions:

• The developed GRNN-based online approach offers an effective solution for uncertain trajectory tracking.
• The novel PPC technique removes the need for strict initial condition constraints, enhancing practical applicability.
• The control scheme demonstrates robustness against unmatched uncertainties and unknown control directions.