An online GA-based output-feedback direct adaptive fuzzy-neural controller for uncertain nonlinear systems

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Computational Intelligence

Background:

• Uncertain nonlinear dynamical systems pose significant control challenges.
• Traditional control methods often struggle with complex system dynamics and uncertainties.
• Adaptive fuzzy-neural controllers offer potential for robust control but require effective tuning mechanisms.

Purpose of the Study:

• To propose a novel Genetic Algorithm (GA)-based output-feedback direct adaptive fuzzy-neural controller (GODAF controller).
• To enhance the online tuning of weighting factors for fuzzy-neural networks in adaptive controllers.
• To ensure the stability of nonlinear systems under adaptive control.

Main Methods:

• Development of a GA-based adaptive fuzzy-neural controller (GODAF controller).
• Utilizing a reduced-form genetic algorithm (RGA) with a sequential-search-based crossover point (SSCP) method for efficient weight tuning.
• Establishing a new fitness function based on the Lyapunov design approach for online tuning.
• Incorporating a supervisory controller to guarantee closed-loop stability.

Main Results:

• The proposed GODAF controller effectively tunes weighting factors online using GA.
• The RGA with SSCP method improves the speed of searching for optimal fuzzy-neural network weights.
• The Lyapunov-based fitness function facilitates stable online tuning.
• Demonstrated effectiveness of the GODAF controller on nonlinear system examples.

Conclusions:

• The novel GODAF controller provides an effective solution for controlling uncertain nonlinear dynamical systems.
• The GA-based online tuning mechanism significantly improves controller performance and adaptability.
• The integration of a supervisory controller ensures system stability, validating the proposed approach.