A1 Journal article (refereed), original research

Deformation modeling of manipulators for DEMO using artificial neural networks


Publication Details
Authors: Li Ming, Wu Huapeng, Handroos Heikki, Skilton Robert, Hekmatmanesh Amin, Loving Antony
Publisher: Elsevier
Publication year: 2019
Language: English
Related Journal or Series Information: Fusion Engineering and Design
Volume number: 146
Start page: 2401
End page: 2406
Number of pages: 6
ISSN: 0920-3796
eISSN: 1873-7196
JUFO-Level of this publication: 1
Open Access: Not an Open Access publication

Abstract

A hybrid deformation modeling method is proposed to model deformation physics of manipulators used in DEMO. The deformation of joints is modeled by neural networks and the hybrid deformation model is constructed by integrating the joint’s deformation with the kinematics of the manipulator. The Markov Chain Monte Carlo Method is employed to identify weight parameters of the artificial neural network. A complex joint assembly of a boom used for JET maintenance is taken as an example of applying the proposed method, which is treated as a single degree of freedom mechanism. The finite element method is used to generate training data for the hybrid model, as well as being used as a benchmark for verifying the trained model. The comparison results indicate that the hybrid modeling method is competent to model the deformation physics of mechanisms assembly, which is kinematically dependent, under force payload.


Last updated on 2020-20-03 at 10:03