Kybernetika 59 no. 1, 28-44, 2023

Performance analysis of least squares algorithm for multivariable stochastic systems

Ziming Wang, Yiming Xing and Xinghua ZhuDOI: 10.14736/kyb-2023-1-0028

Abstract:

In this paper, we consider the parameter estimation problem for the multivariable system. A~recursive least squares algorithm is studied by minimizing the accumulative prediction error. By employing the stochastic Lyapunov function and the martingale estimate methods, we provide the weakest possible data conditions for convergence analysis. The upper bound of accumulative regret is also provided. Various simulation examples are given, and the results demonstrate that the convergence rate of the algorithm depends on the parameter dimension and output dimension.

Keywords:

least squares, martingale theory, non-persistent excitation

Classification:

93A10, 93E12, 93E24

paper.pdf

References:

  1. H.-F. Chen and L. Guo: Strong consistency of recursive identification by no use of persistent excitation condition. Acta Math. Appl. Sinica 2 (1985), 2, 133-145.   DOI:10.1007/BF01539484
  2. H.-F. Chen and L. Guo: Continuous-time stochastic adaptive tracking—robustness and asymptotic properties. SIAM J. Control Optim. 28 (1990), 3, 513-527.   DOI:10.1137/0328030
  3. H.-F. Chen and L. Guo: Identification and Stochastic Adaptive Control. Volume 5. Springer Science Business Media, 1991.   CrossRef
  4. J. C. Doyle, B. A. Francis and A. R. Tannenbaum: Feedback Control Theory. Courier Corporation, 2013.   CrossRef
  5. R. Durrett: Probability: Theory and Examples. Vol. 49. Cambridge University Press, 2019.   CrossRef
  6. D. Gan and Z. X. Liu: On the stability of Kalman filter with random coefficients. IFAC-PapersOnLine 53 (2020), 2, 2397-2402.   DOI:10.1016/j.ifacol.2020.12.2514
  7. L. Guo: Stability of recursive stochastic tracking algorithms. SIAM J. Control Optim. 32 (1994), 5, 1195-1225, 1994.   DOI:10.1137/S0363012992225606
  8. L. Guo: Convergence and logarithm laws of self-tuning regulators. Automatica 31 (1995), 3, 435-450.   DOI:10.1016/0005-1098(94)00127-5
  9. L. Guo: Time-Varying Stochastic Systems, Stability and Adaptive Theory. Second Edition. Science Press, Beijing 2020.   CrossRef
  10. L. Guo, L. Ljung and P. Priouret: Performance analysis of the forgetting factor rls algorithm. Int. J. Adaptive Control Signal Process. 7 (1993), 6, 525-537.   DOI:doi.org/10.1002/acs.4480070604
  11. S. Haykin: Radar signal processing. IEEE Signal Process. Magazine 2 (1993), 2, 2-18.   CrossRef
  12. T. L. Lai: Asymptotically efficient adaptive control in stochastic regression models. Advances Appl. Math. 7 (1986), 1, 23-45.   CrossRef
  13. T. L. Lai and Ch. Z. Wei: Least squares estimates in stochastic regression models with applications to identification and control of dynamic systems. Ann. Statist. 10 (1982), 1, 154-166.   DOI:10.1214/aos/1176345697
  14. J. Li, F. Ding and G. Yang: Maximum likelihood least squares identification method for input nonlinear finite impulse response moving average systems. Math. Comput. Modell. 55 (2012), 3-4, :442-450.   CrossRef
  15. J. Li and P. Stoica: MIMO Radar Signal Processing. John Wiley and Sons, 2008.   CrossRef
  16. Y. Liu and F. Ding: Convergence properties of the least squares estimation algorithm for multivariable systems. Applied Math. Modell. 37 (2013), 1-2, 476-483.   DOI:10.1016/j.apm.2012.03.007
  17. D. Marelli and M. Fu: A continuous-time linear system identification method for slowly sampled data. IEEE Trans. Signal Process. 58 (2010), 5, 2521-2533.   DOI:10.1109/TSP.2009.2040017
  18. J. B. Moore: On strong consistency of least squares identification algorithms. Automatica 14 (1978), 5, 505-509.   DOI:10.1016/0005-1098(78)90010-9
  19. M. Niedzwiecki and L. Guo: Nonasymptotic results for finite-memory wls filters. In: Proc. 28th IEEE Conference on Decision and Control, Vol. 2, 1989, pp. 1785-1790.   CrossRef
  20. F. S. G. Richards: A method of maximum-likelihood estimation. J. Roy. Statist. Soc,: Series B (Methodological) 23 (1961), 2, 469-475.   DOI:10.1111/j.2517-6161.1961.tb00430.x
  21. A. Sen and N. K. Sinha: On-line estimation of the parameters of a multivariable system using matrix pseudo-inverse. Int. J. Systems Sci. 7 (1976), 4, 461-471.   DOI:10.1080/00207727608941930
  22. B. Subudhi and D. Jena: Nonlinear system identification of a twin rotor mimo system. In: IEEE Region 10 Conference 2009, pp. 1-6.   CrossRef
  23. M. Vaezi and A. Izadian: Piecewise affine system identification of a hydraulic wind power transfer system. IEEE Trans. Control Systems Technol. 23 (2015), 6, :2077-2086.   DOI:10.1109/TCST.2015.2398311
  24. W. Wang, F. Ding and J. Dai: Maximum likelihood least squares identification for systems with autoregressive moving average noise. Appl. Math. Modell. 36 (2012), 5, 1842-1853.   DOI:10.1016/j.apm.2011.07.083
  25. T. P. Vogels, K. Rajan and L. F. Abbott: Neural network dynamics. Ann. Rev. Neurosci. 28 (2005), 357-376.   DOI:10.1146/annurev.neuro.28.061604.135637
  26. B. Widrow and S. D. Stearns: Adaptive Signal Processing. Prentice-Hall Englewood Cliffs, NJ 1985.   CrossRef
  27. Y. Zhang: Unbiased identification of a class of multi-input single-output systems with correlated disturbances using bias compensation methods. Math. Computer Modell. 53 (2011), 9-10, 1810-1819.   DOI:10.1016/j.mcm.2010.12.059
  28. Y. Zhang and G. Cui: Bias compensation methods for stochastic systems with colored noise. Appl. Math- Modell. 35 (2011), 4, 1709-1716.   DOI:10.1016/j.apm.2010.10.003