Fast finite-time stabilizing for pure-feedback stochastic nonlinear systems: a neural network dynamic event-triggered strategy

For pure-feedback stochastic nonlinear systems subject to asymmetric constraints, the current nonlinear term is constrained by linear, homogeneous growth conditions. In this work, an entirely new approach to the solution is proposed that removes completely these constraints. The approach begins with the formulation of a Barrier Lyapunov Function (BLF) that is predicated on the constrained states, thereby ensuring that the system’s state variables are maintained within the designated constraints. Subsequently, neural network techniques are employed to address unmodeled dynamics and stochastic disturbances, without imposing any growth conditions. By applying bounded command filtering technique, the analytical computation of the command signal derivatives is avoided successfully. The introduction of a dynamic event-triggered mechanism (DETM), with threshold parameters adjusted in real-time, serves to guarantee the boundedness of all signals within the closed-loop system and enables the system output to track accurately a predefined signal within a finite-time. Finally, numerical and robot manipulator system simulations are offered to augment and validate our theoretical analysis results.