Gain Tuning Using Neural Network for Contact Force Control of Flexible Arm

  • Minoru Sasaki
  • Nobuto Honda
  • Waweru Njeri
  • Kojiro Matsushita

Abstract

In this research, contact force control of an one linkflexible arm is presented. A simple boundary feedback controllerconsisting of bending moment at the base of the flexible arm proposedby Endo et al. A gain adjustment control system using a neuralnetwork is designed and its control performance examined andcompared by numerical simulation and experiment. In this study, wedesigned the feedback gain to correspond to the coupling coefficientof the neural network, and stabilized the learning by giving theinitial value to the coupling coefficient of the neural network, therebyshortening the learning time. Also, in order to adjust the gain value inreal time, a sequential correction type (online learning) that repeatslearning at every sampling was adopted as the learning scheme ofthe neural network. As a result, it was confirmed that by usingthe using the neural network, the value of the feedback gain isadaptively changed and the target contact force converges around0.35 seconds. Comparing with the fixed gain results, it takes shortertime for convergence to the target value by 0.8 seconds, the proposedcontroller is confirmed to be more effective for the contact forcecontrol of the flexible arm.
Published
Feb 19, 2020
How to Cite
SASAKI, Minoru et al. Gain Tuning Using Neural Network for Contact Force Control of Flexible Arm. Proceedings of Sustainable Research and Innovation Conference, [S.l.], p. 241-245, feb. 2020. ISSN 2079-6226. Available at: <http://sri.jkuat.ac.ke/ojs/index.php/proceedings/article/view/887>. Date accessed: 06 apr. 2020.