Combining Integral Concurrent Learning and Adaptive Super-Twisting Control for Unknown Dynamics and External Disturbances

Neural network-based control methods are effective for handling uncertain nonlinear systems by approximating unknown dynamics. However, their reliance on Persistent Excitation (PE) conditions, slow convergence, and susceptibility to chattering under high-frequency disturbances limit practical performance. To address these challenges, this paper presents an innovative control strategy that integrates integral concurrent learning (ICL) with adaptive super-twisting sliding mode control. Radial Basis Function Neural Networks (RBFNNs) are utilized to approximate unknown dynamics online, while ICL leverages both historical and real-time data to improve parameter learning efficiency and eliminate the need for PE conditions. To further enhance robustness, an adaptive super-twisting sliding mode controller with dynamic gain tuning is introduced, ensuring effective disturbance rejection, reduced chattering, and high tracking precision. The combined approach achieves efficient approximation, real-time compensation for external disturbances, and superior sliding motion performance. Stability is rigorously analyzed using Lyapunov theory, and simulations confirm the proposed method’s effectiveness in achieving precise, robust, and adaptive control under uncertain and disturbed conditions.