Search

Your search keyword '"Zhang, Maoguang"' showing total 2 results
Start Over Author "Zhang, Maoguang" Topic algorithms
2 results on '"Zhang, Maoguang"'

1. Reinforcement learning and adaptive optimization of a class of Markov jump systems with completely unknown dynamic information.

Author

He, Shuping, Zhang, Maoguang, Fang, Haiyang, Liu, Fei, Luan, Xiaoli, and Ding, Zhengtao

Subjects
*REINFORCEMENT learning, *ALGORITHMS, *RICCATI equation, *ALGEBRAIC equations, *ADAPTIVE control systems
Abstract

In this paper, an online adaptive optimal control problem of a class of continuous-time Markov jump linear systems (MJLSs) is investigated by using a parallel reinforcement learning (RL) algorithm with completely unknown dynamics. Before collecting and learning the subsystems information of states and inputs, the exploration noise is firstly added to describe the actual control input. Then, a novel parallel RL algorithm is used to parallelly compute the corresponding N coupled algebraic Riccati equations by online learning. By this algorithm, we will not need to know the dynamic information of the MJLSs. The convergence of the proposed algorithm is also proved. Finally, the effectiveness and applicability of this novel algorithm is illustrated by two simulation examples. [ABSTRACT FROM AUTHOR]

Published
2020
Full Text
View/download PDF

2. Online policy iterative-based H∞ optimization algorithm for a class of nonlinear systems.

Author

He, Shuping, Fang, Haiyang, Zhang, Maoguang, Liu, Fei, Luan, Xiaoli, and Ding, Zhengdao

Subjects
*MATHEMATICAL optimization, *ALGORITHMS, *NONLINEAR systems, *RICCATI equation, *OPTIMAL control theory, *ARTIFICIAL neural networks
Abstract

• A new policy iterative algorithm is studied to design the online H ∞ optimization problems for a class of nonlinear systems. • Without considering knowledge regarding the system dynamics, the PI algorithm is derived to design the relevant H ∞ optimal control law by means of policy evaluation and policy improvement related to the corresponding algebraic Riccati equations. • It also proves the convergence of the novel PI algorithm and two simulation results are given to demonstrate the feasibility and the applicability of the designed algorithms. A novel policy iterative scheme for the design of online H ∞ optimal laws for a class of nonlinear systems is presented. First, neural network-based linear differential inclusion techniques with two multi-layered perceptions are applied to linearize the nonlinear terms. Then, an online partially model-free policy iterative scheme is applied to the linearized system to obtain the design the H ∞ optimal control law. The iterative scheme for the linear H ∞ control problem consists of policy evaluation and policy improvement by means of algebraic Riccati equations. We establish the convergence of the novel policy iterative scheme to the optimal control law. Numerical simulations demonstrating the feasibility and applicability of our design algorithm are provided. [ABSTRACT FROM AUTHOR]

Published
2019
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results