LE QUY DON
Technical University
VietnameseClear Cookie - decide language by browser settings

Identification Random Disturbances of Optomechanical Control Systems Based Neural Observers

Van Lanh, N. and Belov, M.P. and Thanh, N.D. (2020) Identification Random Disturbances of Optomechanical Control Systems Based Neural Observers. In: 2020 IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering, EIConRus 2020, 27 January 2020 through 30 January 2020, St. Petersburg and Moscow.

Text
Identification Random Disturbances of Optomechanical Control Systems Based Neural Observers.pdf

Download (485kB) | Preview

Abstract

This paper deals with the application of neural observer to estimate external random disturbance in the form of wind gust dynamic acting on the opto-mechanical control systems. First, we introduce the mathematical model of wind gust disturbances, which are considered as random processes with zero mean. The next section investigates architecture of neural observer based on dynamic recurrent neural networks (DRNN). The learning process of DRNN is described and the performances mean square error (MSE) of different learning algorithms are compared with the aim of studying quality identification and selection of the best learning algorithm for synthesis neural observer. In this work, designing, training, testing of DRNN is carried out in Matlab/ SIMULINK environment. © 2020 IEEE.

Item Type: Conference or Workshop Item (Paper)
Divisions: Faculties > Faculty of Control Engineering
Identification Number: 10.1109/EIConRus49466.2020.9039524
Uncontrolled Keywords: Dynamics; Learning algorithms; Mean square error; Optomechanics; Random processes; Recurrent neural networks; Dynamic recurrent neural networks; Learning process; MATLAB/ SIMULINK; Neural observer; Optomechanical; Quality identifications; Random disturbances; Wind gust; Identification (control systems)
Additional Information: Conference code: 158636. Language of original document: English.
URI: http://eprints.lqdtu.edu.vn/id/eprint/9157

Actions (login required)

View Item
View Item