Self-organizing feature selection fuzzy neural network-based terminal sliding mode control for uncertain nonlinear...

Area of Science:

• Control Engineering
• Artificial Intelligence
• Nonlinear Systems

Background:

• Uncertain nonlinear systems (UNS) present significant control challenges due to unknown or unmeasurable parameters.
• Traditional controllers often struggle with parameter variations, impacting performance and stability.
• Sliding mode control (SMC) offers robustness but can be sensitive to parameter uncertainties.

Purpose of the Study:

• To develop an advanced composite terminal sliding mode controller (CTSMC) for uncertain nonlinear systems (UNS).
• To enhance the control performance of CTSMC by integrating a novel fuzzy neural network (FNN) for adaptive parameter learning.
• To address the challenge of unmeasurable system parameters in real-world UNS applications.

Main Methods:

• Demonstration of the stability and convergence of the CTSMC for UNS with known parameters.
• Development of a self-organizing feature selection fuzzy neural network (SOFSFNN) to approximate unknown system parameters.
• Integration of the SOFSFNN with the CTSMC for adaptive control of UNS.

Main Results:

• The proposed CTSMC with SOFSFNN achieves minimal tracking error in uncertain nonlinear systems.
• The controller exhibits significant robustness against system uncertainties and parameter variations.
• The SOFSFNN demonstrates an ability to dynamically adjust its network structure for improved learning.

Conclusions:

• The developed CTSMC, enhanced by the SOFSFNN, provides a robust and effective solution for controlling uncertain nonlinear systems.
• Adaptive learning of unknown parameters via the SOFSFNN is crucial for achieving high control performance.
• The controller's dynamic network modification capability offers adaptability in complex, real-world scenarios.