Event-Triggered Optimal Tracking Control for Uncertain Nonlinear System Based on Reinforcement Learning

Yuanhao Wang, Weiwei Bai

TL;DR: adequate

Abstract:

In this paper, an event-triggered optimal tracking
control problem is studied for uncertain nonlinear system based
on reinforcement learning (RL). Firstly, a class of nonlinear
dynamic system with general uncertainty is considered and
the augmented system comprising tracking error and reference signal is constructed. Secondly, an improved adaptive
dynamic programming (ADP) technique, involving actor-critic
algorithm and fuzzy logic system, is developed to solve the
Hamilton–Jacobi–Bellman (HJB) equation respect to nominal
augmented system. Thirdly, in order to reduce the mechanical
wear of actuator and energy consumption, event-triggered mechanism is performed in controller update. Finally, stability analysis
proofs the uniformly ultimately bounded (UUB) of all signals in
the closed-loop system via Lyapunov theory. Simulation results
verify feasibility of proposed scheme.

Submission Number: 121