Friction modelling and compensation for motion control using hybrid neural network models

Authors:
M. Kemal Cılız;Masayoshi Tomizuka
Affiliations:
Electrical Engineering Department, Boğaziçi University, Bebek, Istanbul 34342, Turkey;Department of Mechanical Engineering, University of California, Berkeley, CA 94720, USA
Venue:
Engineering Applications of Artificial Intelligence
Year:
2007

Citing 8
Cited 1

Multilayer feedforward networks are universal approximators

Neural Networks
Adaptive friction compensation in robot manipulators: low velocities

International Journal of Robotics Research
A survey of models, analysis tools and compensation methods for the control of machines with friction

Automatica (Journal of IFAC)
Neural Network Control of Robot Manipulators and Nonlinear Systems

Neural Network Control of Robot Manipulators and Nonlinear Systems
On-line learning control of manipulators based on artificial neural network models

Robotica
Adaptive control of robot manipulators with neural network based compensation of frictional uncertainties

Robotica
Adaptive friction compensation using neural network approximations

IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews
Technical Communique: Variable structure control of systems with uncertain nonlinear friction

Automatica (Journal of IFAC)

Support vector regression based friction modeling and compensation in motion control system

Engineering Applications of Artificial Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper investigates artificial neural network (ANN) based modelling and compensation of nonlinear friction which is a major cause of performance degradation in servo mechanisms. Different friction modelling and compensation schemes have been reviewed and neural network based hybrid compensation methods are proposed and experimentally tested on a direct drive servo mechanism. Inertial dynamics is assumed to be constant and a PD type control is deployed for the servo feedback without the motor's electrical dynamics. ANN based techniques resulted in good performance when compared with the experimentally obtained friction model and parametric adaptive models. Advantages and the implementation aspects of the proposed methods are also discussed.