An Online Deep Reinforcement Learning Based Parameter Identification Method for HVDC System

Author(s):  
Jianxiong Hu ◽  
Qi Wang ◽  
Yi Tang
AIP Advances ◽  
2021 ◽  
Vol 11 (5) ◽  
pp. 055302
Author(s):  
Yong Zhu ◽  
Guangpeng Li ◽  
Shengnan Tang ◽  
Wanlu Jiang ◽  
Zhijian Zheng

2019 ◽  
Vol 91 (8) ◽  
pp. 1147-1155 ◽  
Author(s):  
Xiaofeng Liu ◽  
Bangzhao Zhou ◽  
Boyang Xiao ◽  
Guoping Cai

Purpose The purpose of this paper is to present a method to obtain the inertia parameter of a captured unknown space target. Design/methodology/approach An inertia parameter identification method is proposed in the post-capture scenario in this paper. This method is to resolve parameter identification with two steps: coarse estimation and precise estimation. In the coarse estimation step, all the robot arms are fixed and inertia tensor of the combined system is first calculated by the angular momentum conservation equation of the system. Then, inertia parameters of the unknown target are estimated using the least square method. Second, in the precise estimation step, the robot arms are controlled to move and then inertia parameters are once again estimated by optimization method. In the process of optimization, the coarse estimation results are used as an initial value. Findings Numerical simulation results prove that the method presented in this paper is effective for identifying the inertia parameter of a captured unknown target. Practical implications The presented method can also be applied to identify the inertia parameter of space robot. Originality/value In the classic momentum-based identification method, the linear momentum and angular momentum of system, both considered to be conserved, are used to identify the parameter of system. If the elliptical orbit in space is considered, the conservation of linear momentum is wrong. In this paper, an identification based on the conservation of angular momentum and dynamics is presented. Compared with the classic momentum-based method, this method can get a more accurate identification result.


Sign in / Sign up

Export Citation Format

Share Document