<p>This paper investigates a hierarchical sliding mode surface (HSMS)-based dynamic event-triggered optimal control problem for nonlinear systems subject to unknown disturbances. First, a generalized fuzzy hyperbolic model-based identifier network is employed to approximate the unknown nonlinear dynamics. Concurrently, a disturbance observer is constructed to estimate the unknown external disturbances, with the estimated values being systematically incorporated into the cost function for online updating. By formulating a special cost function associated with HSMS, the control problem is transformed into a sequential optimal control strategy derivation task. The Hamilton–Jacobi–Bellman (HJB) equation is then approximately solved through a single-critic neural network architecture enhanced with experience replay (ER) technology. Moreover, a dynamic event-triggered mechanism featuring adaptive threshold adjustment is proposed to optimize communication resource utilization. Through rigorous Lyapunov stability analysis, all closed-loop system signals are proven to be uniformly ultimately bounded (UUB). Finally, the effectiveness of the proposed control scheme is verified by simulation results of a robot arm system.</p>