Science in China Series F: Information Sciences

, Volume 52, Issue 10, pp 1739–1745 | Cite as

Simple recursive algorithm for linear-in-theparameters nonlinear model identification

Special Focus
First Online:
Received:
Accepted:

Abstract

This paper introduces a simple recursive algorithm for nonlinear dynamic system identification using linear-in-the-parameters models for NARX or RBF network where both the structure and parameters can be obtained simultaneously and recursively. The main objective is to improve the numerical stability when the model terms are highly correlated. This is based on the “innovation” idea and net contribution criteria. Using the recursive formulae for the computation of the Moore-Penrose inverse of matrices and the net contribution of model terms, it is possible to combine the structure term determination and parameters estimation within one framework by adding and deleting an item in the selected candidate model. The formulae for enhancing and reducing a matrix are given. Simulation results show the proposed method is numerically more stable than existing approaches.

Keywords

nonlinear dynamic system linear-in-the-parameters models NARX RBF Moore-Penrose inverse 
This is a preview of subscription content, log in to check access.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Haber R, Unbehauen H. Structure identification of nonlinear dynamic systems-a survey on input/output approaches. Automatica, 1990, 26: 651–667MATHCrossRefMathSciNetGoogle Scholar
  2. 2.
    Chen S, Billings S A, Luo W. Orthogonal least squares methods and their application to nonlinear system identification. Int J Control, 1989, 50: 1873–1896MATHCrossRefMathSciNetGoogle Scholar
  3. 3.
    Brouwn G G, Krijgsman A J. Single-layer networks for nonlinear system identification. Eng Appl Artif Intel, 1994, 7: 227–243CrossRefGoogle Scholar
  4. 4.
    Li K, Peng J X, Irwin G W. A fast nonlinear model identification method. IEEE Trans Automat Control, 2005, 50: 1211–1216CrossRefMathSciNetGoogle Scholar
  5. 5.
    Chen S, Cowan F N, Grant P M. Orthogonal least squares learning algorithm for radial basis functions. IEEE Trans Neural Netw, 1991, 2: 302–309CrossRefGoogle Scholar
  6. 6.
    Hong X, Billings S A. Givens rotation based fast backward elimination algorithm for rbf neural network pruning. IEE Proc D, Control Theory Appl, 1997, 144: 381–384MATHCrossRefGoogle Scholar
  7. 7.
    Huang G B, Saratchandran P, Sundararajan N. A generalized growing and pruning rbf (ggap-rbf) neural network for function approximation. IEEE Trans Neural Netw, 2005, 16: 57–67CrossRefGoogle Scholar
  8. 8.
    Vogt M. Combination of radial basis function neural networks with optimized learning vector quantization. Proc IEEE, 1993, 83: 1841–1846Google Scholar
  9. 9.
    Li K, Peng J, Bai E W. A two-stage algorithm for identification of nonlinear dynamic systems. Automatica, 2006, 42(7): 1189–1197MATHCrossRefMathSciNetGoogle Scholar
  10. 10.
    Peng J, Li K, Huang D S. A hybrid forward algorithm for RBF neural network construction. IEEE Trans Neural Netw, 2006, 17(6): 1439–1451CrossRefGoogle Scholar
  11. 11.
    Huang D S. The united adaptive learning algorithm for the link weights and the shape parameters in rbfn for pattern recognition. Int J Pattern Recogn, 1997, 11(6): 873–888CrossRefGoogle Scholar
  12. 12.
    Chen S. Billings S A. Neural network for nonlinear dynamic system modelling and identification. Int J Control, 1992, 56: 319–346MATHCrossRefMathSciNetGoogle Scholar
  13. 13.
    Zhu Q M, Billings S A. Fast orthogonal identification of nonlinear stochastic models and radial basis function neural networks. Int J Control, 1996, 64(5): 871–886MATHCrossRefMathSciNetGoogle Scholar
  14. 14.
    Hong X, Harris C J, Chen S, et al. Robust nonlinear model identification methods using forward regression. IEEE Trans Syst Man Cybern A, 2003, 33(4): 514–523CrossRefGoogle Scholar
  15. 15.
    Chen S, Billings S A, Luo W. Orthogonal least squares methods and their application to non-linear system identification. Int J Control, 1989, 50(5): 1873–1896MATHCrossRefMathSciNetGoogle Scholar
  16. 16.
    Phohomsiri P, Han B. An alternative proof for the recursive formulae for computing the moore-penrose m-inverse of a matrix. Appli Math Comput, 2006, 174: 81–97MATHCrossRefMathSciNetGoogle Scholar
  17. 17.
    Saleem M, Vladimer C. On recursive calculation of the generalized inverse of a matrix. ACM Trans Math Softw, 1994, 17: 130–147Google Scholar
  18. 18.
    Xiufeng W, Yuhong L. A new algorithm for model structure determination and parameters estimation of nonliner dynamic systems. Acta Automat Atica Sin, 1992, 18: 385–392MATHGoogle Scholar
  19. 19.
    Papaodysseus C, Roussopoulos A P. Using a fast RLS adaptive algorithm for efficient speech processing. Math Comput Simulat, 2005, 68: 105–113MATHCrossRefMathSciNetGoogle Scholar
  20. 20.
    Djigan V I. Multichannel parallelizable sliding window RLS and fast RLS algorithms with linear constraints. Signal Process, 2006, 86: 776–791MATHCrossRefGoogle Scholar
  21. 21.
    Li P K, Kruger U, Du X X. A fast noise filtering approach using HILS for electrical vortex ergograph system. In: Proceedings of IEEE 7th International Conference on Signal Processing. Vol. 1. Beijing, China. New York: IEEE Press, 2004. 575–578Google Scholar
  22. 22.
    Mackey M C, Glass L. Oscillation and chaos in physiological control systems. Science, 1977, 197: 287–289CrossRefGoogle Scholar
  23. 23.
    Cho K B, Wang B H. Radial basis function based adaptive fuzzy systems and their applications to system identification and prediction. Fuzzy Sets Syst, 1996, 83(3): 325–339CrossRefMathSciNetGoogle Scholar

Copyright information

© Science in China Press and Springer Berlin Heidelberg 2009

Authors and Affiliations

  1. 1.School of Mechanical, Electrical and Control EngineeringBeijing Jiaotong UniversityBeijingChina

Personalised recommendations