Off-Policy Reinforcement Learning for Optimal Preview Tracking Control of Linear Discrete-Time systems with unknown dynamics

摘要

In this paper, an off-policy reinforcement learning (RL) algorithm is presented to solve the optimal preview tracking control of discrete time systems with unknown dynamics. Firstly, an augmented state-space system that includes the available preview knowledge as a part of the state vector is constructed to cast the preview tracking control problem as a standard linear quadratic regulator (LQR) one. Secondly, the reinforcement learning technique is utilized to solve the algebraic Riccati equation (ARE) using online measurable data without requiring the a priori knowledge of the system matrices. Compared with the existing off-policy RL algorithm, the proposed scheme solves a preview tracking control problem. A numerical simulation example is given to verify the effectiveness of the proposed control scheme.