Inverse updated systolic RLS algorithm with regularized exponential forgetting. (English) Zbl 0870.93051

Schier, Jan

Kybernetika 32, No. 3, 209-234 (1996).

This paper investigates the implementation of the idea of block-regularized exponential forgetting in the systolic algorithm for the recursive least squares (RLS) identification. It combines the inverse updated systolic RLS algorithm with covariance update and the block-accumulated regularized exponential forgetting. The regularized forgetting prevents the covariance matrix from unlimited growth and then increases numerical stability of the algorithm for weakly informative data. It is shown that the proposed implementation of the block-regularized forgetting preserves compactness of the systolic estimator with exponential forgetting since the algorithm uses only the connections between the neighboring processors.

Reviewer: K.Uosaki (Tottori)

MSC:

93E24	Least squares and related methods for stochastic control systems
93E12	Identification in stochastic control theory

Keywords:

recursive least squares algorithm; block regularized exponential forgetting; systolic algorithm; stability

References:

[1]	D. D. Baer J. Paradeans: A formal definition for systolic systems. Parallel Algorithms and Architectures (A. Albrecht, H. Jung and K. Mehlhorn, Lecture Notes in Computer Science, Springer-Verlag, Berlin 1987. · Zbl 0634.68013
[2]	L. D. J. Eggermont, al. (eds.): VLSI Signal Processing VI. Proceedings of the IEEE Signal Processing Society Workshop. IEEE Press, Veldhoven 1993.
[3]	J. Kadlec: The cell-level description of systolic block regularised QR filter. Proceedings of the IEEE Signal Processing Society Workshop (Eggermont et al., Veldhoven 1993, pp. 298-306.
[4]	J. Kadlec F. M. F. Gaston G. W. Irwin: Systolic implementation of the regularised parameter estimator. VLSI Signal Processing V (K. Yao et al., IEEE Press, New York 1992, pp. 520-529.
[5]	M. Kárný, al.: Design of linear quadratic adaptive control: Theory and algorithms for practice. Kybernetika 21 (1985), Supplement.
[6]	J. G. McWhirter: Systolic array for recursive least squares by inverse iterations. Proceedings of the IEEE Signal Processing Society Workshop (Eggermont et al., Veldhoven 1993, pp. 435-443.
[7]	J. G. McWhirter: A systolic array for recursive least squares estimation by inverse updates. International Conference on Control ’94, IEE, London 1994.
[8]	G. M. Megson: An Introduction to Systolic Array Design. Oxford University Press, Oxford 1992. · Zbl 0807.68030
[9]	M. Moonen J. G. McWhirter: A systolic array for recursive least squares by inverse updating. Electronics Letters 29 (1993), 13, 1217-1218.
[10]	V. Peterka: Bayesian approach to system identification. Trends and Progress in System Identification (P. Eykhoff, IFAC Series for Graduates, Research Workers and Practising Engineers, Chapter 8. Pergamon Press, Oxford 1981. · Zbl 0451.93059
[11]	V. Peterka: Control of uncertain processes: Applied theory and algorithms. Kybernetika 22 (1986), Supplement. · Zbl 0607.93070
[12]	J. Schier: Parallel Algorithms for Robust Adaptive Identification and Square-root LQG Control. Ph.D. Thesis, Faculty of Nuclear Sciences and Physical Engineering, Czech Technical University, Prague 1994.
[13]	J. Schier: A Systolic Algorithm for the Block-regularized RLS Identification. Research Report No. 1807, Institute of Information Theory and Automation, Prague 1994.