Abstract
For high accuracy in mechanical control, the compensation of friction must be taken into account When working at low velocities, the slip-stick phenomenon appears, introducing additive perturbation torque. Mechanical and robotics engineers use several models to simulate these torques. Here we select two dynamic friction laws (Dahl model and Reset-Integrator model), and we use Recurrent Neural Networks to model these dynamic systems in order to complete the «a priori» knowledge with what we don't know how to modelize. We use the «canonical» architecture to construct our network, and use a gradient based algorithm to train it. Results show that a general architecture for describing this family of friction laws can be obtained. This type of architecture may be used in regulation schemes.
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
References
J. Gilbart, G. Winston (1974): «Adaptative Compensation for an Optical Tracking Telescope» Automatica Vol 10, pp 125–131
C.Walrath (1984): «Adaptative Bearing Friction Compensation Based on Recent Knowledge of Dynamic Friction», Automatica Vol 20, N∘6, pp717–727
«Compensation de frottements secs sur un système de stabilisation 3 axes » SFIM Industries, Internal Report 1992.
R.Murray-Smith, D.Neumerkel (1992): «Neural Networks for Modelling and Control of a Non-Linear Dynamic System» Proc. of the 1992 IEEE Int. Symp. on Intelligent Control, pp404–409.
P.R.Dahl (1968): «A solid friction model» AFO 4695-67-C-0158, Aeropace Corporation, El Segundo, CA
D.Haessig, B.Friedland (1991): «On the Modeling and Simulation of Friction» Trans. of ASME Jrnl. of Dyn. Syst., Meas. and Contr. Vol 113, pp-362.
K.S.Narendra, K.Parthasarathy (1990): «Identification and Control of Dynamical Systems Using Neural Networks» IEEE Trans. on Neural Networks, March 1990, Vol 1, Nb 1, pp 4–27.
O.Nerrand, P.Roussel-Ragot, D.Urbani, L.Personnaz, G.Dreyfus (1994): «Training Recurrent Neural Network: Why and How? An illustration in Dynamical Process Modeling» IEEE Trans. on Neural Networks, March 1994, Vol 5, Nb 2, pp 178–185.
D.E. Rumelhart, G.E.Hinton, R.J.Williams (1986): «Learning internal representations by error propagation» Parallel Distributed Processing: Explorations in the Microstructures of Cognition, Vol. 1, MIT Press.
P.J. Werbos (1990): «Backpropagation Through Time: What It Does and How to Do it» Proceedings of the IEEE, October 1990, vol 78, no 10, pp1550–1560.
J.M. Martinez, C.Parey, M.Houkari (1992): « Learning Optimal Control using Neural Networks» Proceedings of Neuro-Nimes, pp 431–441.
Y. Le Cun (1994): «Efficient Learning and Second Order Methods» Numerical Analysis Summer School, Neural Networks and Applications, June 13th–24th 1994 CEA, INRIA, EDF FRANCE
Leontaritis, Billings, (1985): «Input-output parametric models for non-linear systems », Int. J. Control, vol 41, n∘2, 303–328
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1995 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Dominguez, M., Michelin, J.M., Martinez, J.M. (1995). Recurrent neural networks for identification of friction. In: Mira, J., Sandoval, F. (eds) From Natural to Artificial Neural Computation. IWANN 1995. Lecture Notes in Computer Science, vol 930. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-59497-3_285
Download citation
DOI: https://doi.org/10.1007/3-540-59497-3_285
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-59497-0
Online ISBN: 978-3-540-49288-7
eBook Packages: Springer Book Archive