Composite Learning-Based Adaptive Terminal Sliding Mode Control for Nonlinear Systems With Experimental Validation

In this article, we introduce a novel neural network (NN)-based indirect adaptive terminal sliding mode control (TSMC) approach for enhancing the identification and control accuracy of nonlinear systems while overcoming potential singularity issues. Initially, the original nonlinear system is transformed into a new format to facilitate the implementation of a singularity-free control framework in subsequent stages. Subsequently, an online learning algorithm is developed for estimating unknown parameters and NN weights, ensuring finite-time convergence of weight errors. A TSMC is then designed within this singularity-free control framework to guarantee finite-time convergence of tracking errors while avoiding potential singularities caused by unknown control gains. Additionally, a composite learning algorithm is proposed to further enhance identification and control performance. The closed-loop system's practical finite-time stability is rigorously proved using the Lyapunov approach. Experimental results on a piezoactuator (PEA) system demonstrate the effectiveness of the proposed identification and control algorithms.