Recurrent fuzzy neural network backstepping control for the prescribed output tracking performance of nonlinear...

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Nonlinear Dynamics

Background:

• Traditional backstepping control methods for nonlinear systems often lead to computational complexity due to recursive term explosion.
• Achieving prescribed tracking performance with strict error bounds in nonlinear dynamic systems remains a significant challenge.
• Recurrent Fuzzy Neural Networks (RFNNs) offer potential for improved approximation capabilities in complex control tasks.

Purpose of the Study:

• To propose a novel backstepping control system for strict-feedback nonlinear dynamic systems.
• To ensure prescribed tracking performance by constraining the system's tracking error within predefined boundaries.
• To mitigate the computational complexity associated with traditional backstepping techniques.

Main Methods:

• A backstepping control framework incorporating a novel constraint variable to enforce tracking error bounds.
• Utilization of adaptive Recurrent Fuzzy Neural Networks (RFNNs) to approximate complex system dynamics and avoid recursive term explosion.
• Analysis of closed-loop system stability and convergence using Lyapunov stability theory.

Main Results:

• The proposed control system successfully forces the tracking error within prescribed boundaries.
• Adaptive RFNNs demonstrated improved approximation performance, reducing computational load compared to traditional methods.
• Lyapunov stability theory confirmed the boundedness and convergence of the closed-loop system.

Conclusions:

• The developed backstepping control scheme effectively achieves prescribed performance for nonlinear systems.
• The integration of RFNNs provides an efficient alternative to traditional backstepping, enhancing approximation capabilities.
• The control strategy was validated on a nonlinear system and a robot manipulator, demonstrating its practical applicability.