SciELO - Scientific Electronic Library Online

 
vol.9 issue1Wood Defects Classification Using Artificial Neural NetworkThe Practical Design Method: A Software Design Method for a First Object-Oriented Project author indexsubject indexsearch form
Home Pagealphabetic serial listing  

Services on Demand

Journal

Article

Indicators

Related links

  • Have no similar articlesSimilars in SciELO

Share


Computación y Sistemas

Print version ISSN 1405-5546

Comp. y Sist. vol.9 n.1 México Jul./Sep. 2005

 

Implementación de un Multimodelo Neuronal Jerárquico para Identificación y Control de Sistemas Mecánicos

 

Implementation of a Neural Hierarchical Multimodel for Identification and Control of Mechanical Systems

 

Ieroham Baruch y José Luis Olivares Guzmán

 

CINVESTAV–IPN Departamento de Control Automático, Av. IPN 2508 Col. San Pedro Zacatenco, A.P. 14–740 C.P. 07360, México D.F., México baruch@ctrl.cinvestav.mx ; jolivares@ctrl.cinvestav.mx

 

Artículo recibido en octubre 17, 2003
Aceptado en abril 01, 2005

 

Resumen

En este artículo se propone la implementación de un Multimodelo Neuronal Jerárquico (MNJ) basándose en la símilarridad con el modelo difuso de Takagi–Sugeno. El modelo MNJ tiene tres partes: 1) fuzificación; 2) inferencia en el nivel bajo usando Redes Neuronales Recurrentes, RNR; 3) defuzifición en el nivel jerárquico alto usando una RNR que es en realidad un filtro–sumador ponderado de las salidas de las RNR del nivel bajo. El aprendizaje y el funcionamiento de ambos niveles jerárquicos son independientes. El modelo MNJ es implementado como identificador y controlador (feedforward, y feedback) en dos esquemas de control directo adaptable. Ambos esquemas de control son aplicados con una planta mecánica con fricción y comparados con otros esquemas de control neuronal y difuso, mostrando mejores resultados.

Palabras Clave: Control adaptable neuronal con modelo inverso, control neuronal directo adaptable, identificación de sistemas, Multimodelo Neuronal Jerárquico, Red Neuronal Recurrente Entrenable, sistema mecánico con fricción.

 

Abstract

The present paper proposed to implement a Neural Hierarchical Multi–Model (MNJ) based on the similarity with the fuzzy model of Takagi–Sugeno. The MNJ has three parts: 1) fuzzyfication; 2) inference engine in the lower hierarchical level, using Recurrent Neural Networks, RNR; 3) defuzzyfication in the upper hierarchical level, using one RNR doing a filtered weighted summation of the outputs of the lower level RNRs. The learning and functioning of both hierarchical levels is independent. The MNJ is implemented in two schemes of direct adaptive control as an identifier and as a feedforward/feedback controller, as well. Both control schemes are applied for control of a mechanical plant with friction and compared with other neural and fuzzy control schemes, exhibiting better results.

Keywords: Inverse model adaptive neural control, direct adaptive neural control, systems identification, Neural Hierarchical Multimodel, Recurrent Trainable Neural Network, mechanical system with friction.

 

DESCARGAR ARTÍCULO EN FORMATO PDF

 

Referencias

1. Karnopp, D.: Computer Simulation of Stick–Slip Friction in Mechanical Dynamic Systems. ASME Journal of Dynamic Systems, Measurement, and Control 107(1985) 100–103.        [ Links ]

2. Lee, S.W., Kim, J.H.: Robust Adaptive Stick–Slip Friction Compensation. IEEE Trans. Ind. Electr. 42 (1995) 474–479.        [ Links ]

3. Menon, K., Krihnamurthy, K.: Control of Low Velocity Friction and Gear Backlash in a Machine Tool Feed Drive System. Mechatronics 9 (1999) 33–52.        [ Links ]

4. Baruch, I., Gortcheva, E.: Fuzzy Neural Model for Nonlinear Systems Identification. In: Proc. of the AARTC'98 IF AC Workshop, Cancun, Mexico, 15–17 April, (1998) 283–288.        [ Links ]

5. Baruch, I., Gortcheva, E., Thomas, F., Garrido, R.: A Neuro–Fuzzy Model for Nonlinear Plants Identification. In: Proc. of the IASTED Int. Conf. on Modeling and Simulation, MS'99, Philadelphia, PA, USA, May 5–8, (1999) 1–6.        [ Links ]

6. Baruch, I., Thomas, F., Garrido, R., Gortcheva, E.: A Hybrid Multimodel Neural Network for Nonlinear Systems Identification. In: Proc. of the Int. Joint Conference on Neural Networks, Washington D.C., USA, July 10–16, 6 (1999) 4278–4283.        [ Links ]

7. Baruch, I., Garrido, R., Mitev, A., Nenkova, B.: A Neural Network Approach for Stick–Slip Friction Model Identification. In: Proc. of the 5–th Int. Conf. on Engineering Applications of NNs, Warsaw, Poland, Sept. 13–15, (1999) 183–188.        [ Links ]

8. Baruch, I., Flores, J.M., Garrido, R., Gortcheva, E.: Identificación de Sistemas No Lineales Complejos Usando un Multimodelo Neuronal Difuso. Científica, ESIME, 19 (2000) 29–40.        [ Links ]

9. Baruch, I., Flores, J.M., Thomas, F., Gortcheva, E.: A Multimodel Recurrent Neural Network for Systems Identification and Control. In: Proc. of the International Joint Conference on Neural Networks, Washington D.C., USA, July 14–19, (2001) 1291–1296.        [ Links ]

10. Baruch, I., Flores, J.M., Thomas, F., Garrido, R.: Adaptive Neural Control of Nonlinear Systems. In: Proc. of the Artificial Neural Networks Conf. –ICANN, Lecture Notes in Comp. Science, Vol. 2130, Springer–Verlag, Berlin, Heidelberg, New York (2001) 930–936.        [ Links ]

11. Baruch, I., Flores, J.M., Nava, F., Ramirez, R., Nenkova, B.: An Advanced Neural Network Topology and Learning, Applied for Identification and Control of a D.C. Motor. In: Proc. of the 1–st Int. IEEE Symposium on Intel. Systems, Varna, Bulgaria, Sept., (2002) 289–295.        [ Links ]

12. Ramirez, I.R., Baruch, I., Garrido, R.: Neuro Control Adaptable para un Motor CD. Científica, ESIME, 6, 3, Julio–Septiembre, (2002) 133–142.        [ Links ]

13. Baruch, I., Beltran, R., del Pozo, A., Garrido, R.: Control Multimodelo Neuronal para Sistemas Electromecanicos. Revista Ingeneria Electronica, Automatica y Comunicaciones, ISPJAE, La Habana, Cuba, ISSN 0258–5944, XXV, 1 (2004) 8–17.        [ Links ]

14. Lin Chin–Ten, Lee C.S. George: Neural Fuzzy Sistem, A Neuro–Fuzzy Synergism to Intelliegent Systems. Prentice – Hall PTR, New Jersey (1996).        [ Links ]

15. Teixeira, M., Zak, S: Stabilizing Controller Design for Uncertain Nonlinear Systems Using Fuzzy Models. IEEE Trans. Syst., Man, and Cyb., 7 (1999) 133–142.        [ Links ]

16. Mastorocostas, P.A., Theocharis, J.B.: Recurrent Fuzzy–Neural Model for Dynamic System Identification. IEEE Trans. Syst., Man, and Cyb. – Part B: Cybernetics, 32 (2002) 176–190.        [ Links ]

17. Takagi, T., Sugeno, M.: Fuzzy Identification of Systems and Its Applications to Modeling and Control. In: IEEE Trans. Syst., Man, and Cyb., 15 (1985) 116–132.        [ Links ]

18. Frasconi, P., Gori, M., Soda, G.: Local Feedback Multilayered Networks. Neural Computation, 4 (1992) 120–130.        [ Links ]

19. Narendra, K. S., Parthasarathy, K.: Identification and Control of Dynamical Systems Using Neural Networks. IEEE Trans. Neural Networks, 1 (1990) 4–27.        [ Links ]

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License