Emotional Neural Network-Based Global Predefined-Time Sliding Mode Control for Uncertain Hybrid Mechanism

An emotional neural network-based global predefined-time sliding mode control (ENN-GPTSMC) method is proposed for an uncertain hybrid mechanism. To estimate and compensate for the lumped uncertainty including discontinuous friction, an emotional neural network is developed. Simultaneously, a predefined-time terminal sliding mode control (PTTSMC) uses the estimation value. The adjustable predefined-time performance parameters are then incorporated into the PTTSMC law to extend its attractiveness for the system states to the global domain, thereby solving the limitation of the existing PTTSMC that can only locally achieve the predefined-time convergence of the system states during the reaching phase. The fast convergence of system states is subsequently achieved by embedding an integer-power linear term and its derivative into the sliding manifold and PTTSMC law, respectively. Based on these, an ENN-GPTSMC algorithm is designed. Furthermore, the saturation function of a dynamic boundary layer with an adjustable thickness is designed to avoid the singularity of ENN-GPTSMC, thereby achieving no-singularity fast global predefined-time convergence of the system. Theoretical analysis shows the Lyapunov stability of the system. Finally, simulation and prototype experiments are used to verify the effectiveness of the proposed method.