Event‐Triggered Optimal Control of Nonlinear Systems via Disturbance Observer‐Based Reinforcement Learning Backstepping Approach

This article addresses the event‐triggered optimized backstepping control problem for uncertain nonlinear strict‐feedback systems with mismatched disturbances. By constructing a class of disturbance observers (DOs) to estimate lumped disturbances, neural network (NN)‐based reinforcement learning (RL) is employed within an identifier‐critic‐actor framework to achieve the optimized control. Meanwhile, a novel state‐dependent event‐triggered scheme is delicately designed to reduce communication consumption. Using the backstepping approach, both virtual and actual controllers are optimized while ensuring the system stability via Lyapunov theory and preventing Zeno behavior by ensuring positive inter‐event intervals. The results are validated through simulations on a multi‐manipulator system.