AUTHORS: Hong Son Hoang, Remy Baraille
Download as PDF
ABSTRACT: This paper presents an optimal filtering approach to state and model error (ME) estimation problem, with a deterministic or stochastic ME. The approach is based on the adaptive filtering (AF) algorithm which is aimed at overcoming the difficulties in the filter design with very high dimensionality of the dynamic systems. The objective is to design a filtering algorithm offering potential for improvement of numerical accuracy and reduction of computational burden. A hypothesis on the structure of ME is introduced. The improvement of the AF performance is achieved by tuning some pertinent parameters of the filter gain as well as bias parameters to minimize the prediction error of the system output. Numerical experiments are presented to illustrate the performance of the proposed approach.
KEYWORDS: Dynamic system, Model error, Adaptive filter, Minimal mean prediction error, Filter stability
REFERENCES:
[1] D.P. Dee, On-line estimation of error covariance parameters for atmospheric data assimilation, Mon. Weather Rev., 123, pp. 1995, 1128 - 1145.
[2] M. Dowd, Estimating parameters for a stochastic dynamic marine ecological system, Environmetrics, 22, 2011, pp. 501 - 515.
[3] B. Friedland, Treatment of bias in recursive filtering, IEEE Trans. Automat. Contr., AC14, 1969,pp. 359 - 367.
[4] A.H. Jazwinski, Stochastic Processes and Filtering Theory, New York: Academic, 1970.
[5] G.H. Golub and C.F. Van Loan, 1993. Matrix Computations. 2 edn. Johns Hopkins.
[6] H.S. Hoang and B. Baraille, Prediction error sampling procedure based on dominant Schur decomposition: Application to state estimation in high dimensional oceanic model, J. Appl. Math. Comput. 218, 2011, pp. 3689–3709.
[7] H.S. Hoang and B. Baraille, Dynamical Systems-based Approach for Simulation and Estimation of Random Processes with Separable Covariance Functions, In Mathematical models and methods in modern science, Ed. A. J. Viamonte, 2012, pp. 36-41.
[8] H.S. Hoang and B. Baraille, On the efficient low cost procedure for estimation of highdimensional prediction error covariance matrices Automatica 83, 2017, pp. 317 - 330.
[9] H.S. Hoang and B. Baraille, A simple numerical method based simultaneous stochastic perturbation for estimation of high dimensional matrices, Multidimensional Systems and Signal Processing, 2019, Issue 1, pp 195 - 217.
[10] H.S. Hoang, B. Baraille and O. Talagrand, On an adaptive filter for altimetric data data assimilation and its application to a primitive equation model MICOM, Tellus 57A, 2005, pp. 153-170.
[11] H.S. Hoang, P. DeMey, B. Baraille and O. Talagrand, A new reduced-order adaptive filter for state estimation in high dimensional systems, Automatica 33, 1997, pp. 1475-1498.
[12] H.S. Hoang, O. Talagrand and R. Baraille, On the design of a stable adaptive filter for high dimensional systems. Automatica, 37, 2001, pp. 341-359.
[13] F.-X. Le Dimet and V. Shutyaev. On deterministic error analysis in variational data assimilation. Nonlinear Processes in Geophysics, 12, (4), (2005), p. 481–490. doi:10.5194/npg-12-481- 2005.
[14] J. M. Mendel, Extension of Friedlands bias filtering technique to a class of nonlinear systems, IEEE Trans. Automat. Contr., AC-21, 1976, pp. 296 - 298.
[15] J.C. Spall, Multivariate Stochastic Approximation Using a Simultaneous Perturbation Gradient Approximation”, IEEE Trans. Autom. Contr., vol. 37(3), pp. 332-341, 1992.
[16] E. L. Shreve and W. R. Hedrick, Separating bias and state estimation in a recursive second-order filter, IEEE Trans. Automat. Contr., AC-19, 1974, pp. 585 - 586.
[17] E. C. Tacker and C. C. Lee, Linear filtering in the presence of time-varying bias, IEEE Trans. Automat. Contr., AC-17, 1972, pp. 828 - 829.
[18] D. H. Zhou, Y. X. Sun, Y. G. Xi, and Z. J. Zhang, Extension of Friedlands separate-bias estimation to randomly time-varying bias of nonlinear systems, IEEE Trans. Automat. Contr., vol. 38, 1993, pp. 1270 - 1273.