Adaptive Fixed‐Time Composite Learning Control of Uncertain Robotic Manipulator Systems Based on Fuzzy System and Neural Network

ABSTRACT This article proposes a novel adaptive error‐driven composite learning control (AEDCLC) scheme for robotic manipulator systems. The method is developed based on an adaptive error‐driven composite recurrent neural network (AEDCRNN) and a nonsingular fast terminal sliding mode control (NFTSMC) framework. This combined strategy effectively prevents parameter drift during the learning of lumped uncertainties, while the integration of NFTSMC substantially enhances the robustness of the overall system. Firstly, a NFTSMC scheme is constructed to ensure robust tracking performance, rapid convergence, and minimized control chattering for the robotic manipulator system. Secondly, a novel AEDCRNN algorithm incorporating a fixed‐time adaptive tuning strategy and an error scheduling mechanism, is developed to improve global learning capability and convergence speed. Subsequently, a novel fault‐tolerant controller is devised based on NFTSMC scheme and AEDCRNN. Then, rigorous stability analysis is conducted, guaranteeing globally practical fixed‐time stability of the closed‐loop system. Finally, comparative simulations are performed to demonstrate the effectiveness and improved performance of the proposed approach.