Model-free finite-horizon optimal tracking control of discrete-time linear systems

Highlights：

• In this paper, the finite horizon linear quadratic tracking problem transforms to the finite-horizon linear quadratic regulator problem with the help of the augmented system.

• The developed algorithm in this paper solves the time-varying Riccati equations offline and backward-in-time without the system dynamics or any model identification schemes through the defined time-varying Q-function.

• Compare with the work in [56], the restrictions on initial state of the developed algorithm in this paper are not required. Besides, optimal tracking control can be achieved in this paper.

• Compare with the work in [57], the discrete-time system is studied in this paper. While, continuous-time system is considered in [57]. Besides, there is no restriction on the initial value function of the developed algorithm in this paper, while the initial value function of the developed algorithm in [57] needs to meet certain conditions.

摘要

•In this paper, the finite horizon linear quadratic tracking problem transforms to the finite-horizon linear quadratic regulator problem with the help of the augmented system.•The developed algorithm in this paper solves the time-varying Riccati equations offline and backward-in-time without the system dynamics or any model identification schemes through the defined time-varying Q-function.•Compare with the work in [56], the restrictions on initial state of the developed algorithm in this paper are not required. Besides, optimal tracking control can be achieved in this paper.•Compare with the work in [57], the discrete-time system is studied in this paper. While, continuous-time system is considered in [57]. Besides, there is no restriction on the initial value function of the developed algorithm in this paper, while the initial value function of the developed algorithm in [57] needs to meet certain conditions.