Spectrum estimation with missing values: A regularized nuclear norm minimization approach

In this paper, we consider estimation of a power spectrum from noise corrupted spectrum samples on uniform grids of frequencies with missing values. We propose two schemes based on the regularized nuclear norm minimization in combination with a recent subspace identification algorithm. The proposed schemes estimate the model order and the missing spectrum values in one step and are robust to large amplitude noise over short data records. Although this estimation problem can be cast as a spectrum estimation problem from nonuniformly spaced measurements and the algorithms developed for this type of data can be used, the identification example of this paper shows that the incomplete data formulation yields more accurate results. The properties of one of the proposed schemes are illustrated in an application example concerned with low-pass modeling of transformer current.

[1]  Bingsheng He,et al.  A new inexact alternating directions method for monotone variational inequalities , 2002, Math. Program..

[2]  Lennart Ljung,et al.  System identification toolbox for use with MATLAB , 1988 .

[3]  Richard H. Jones,et al.  Maximum Likelihood Fitting of ARMA Models to Time Series With Missing Observations , 1980 .

[4]  Hüseyin Akçay,et al.  Positive realness in stochastic subspace identification: A regularized and reweighted nuclear norm minimization approach , 2015, 2015 European Control Conference (ECC).

[5]  Huseyin Akcay.  Frequency domain subspace identification of discrete-time singular power spectra , 2011, 2011 International Siberian Conference on Control and Communications (SIBCON).

[6]  Ivan Markovsky,et al.  Structured Low-Rank Approximation with Missing Data , 2013, SIAM J. Matrix Anal. Appl..

[7]  Hüseyin Akçay,et al.  Frequency domain subspace-based identification of discrete-time power spectra from nonuniformly spaced measurements , 2004, Autom..

[8]  Susan A. Murphy,et al.  Linear fitted-Q iteration with multiple reward functions , 2013, J. Mach. Learn. Res..

[9]  Rik Pintelon,et al.  Frequency domain system identification with missing data , 1998, Proceedings of the 37th IEEE Conference on Decision and Control (Cat. No.98CH36171).

[10]  Hüseyin Akçay,et al.  Frequency domain subspace-based identification of discrete-time power spectra from nonuniformly spaced measurements , 2004, 2004 43rd IEEE Conference on Decision and Control (CDC) (IEEE Cat. No.04CH37601).

[11]  Anders Hansson,et al.  Maximum likelihood estimation of Gaussian models with missing data - Eight equivalent formulations , 2012, Autom..

[12]  Lieven Vandenberghe,et al.  Semidefinite programming methods for system realization and identification , 2009, Proceedings of the 48h IEEE Conference on Decision and Control (CDC) held jointly with 2009 28th Chinese Control Conference.

[13]  Paul Tseng,et al.  Hankel Matrix Rank Minimization with Applications to System Identification and Realization , 2013, SIAM J. Matrix Anal. Appl..

[14]  Donald B. Rubin,et al.  Max-imum Likelihood from Incomplete Data , 1972 .

[15]  Stephen P. Boyd,et al.  Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers , 2011, Found. Trends Mach. Learn..

[16]  Hüseyin Akçay,et al.  Spectral estimation in frequency-domain by subspace techniques , 2014, Signal Process..

[17]  Maryam Fazel,et al.  Iterative reweighted algorithms for matrix rank minimization , 2012, J. Mach. Learn. Res..

[18]  D. Rubin,et al.  Statistical Analysis with Missing Data , 1988 .

[19]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[20]  Stanley Osher,et al.  A Unified Primal-Dual Algorithm Framework Based on Bregman Iteration , 2010, J. Sci. Comput..

[21]  E. Ab.  A subspace method for frequency selective identification of stochastic systems , 2008 .

[22]  Zhang Liu,et al.  Nuclear norm system identification with missing inputs and outputs , 2013, Syst. Control. Lett..

[23]  Hüseyin Akçay,et al.  Nuclear Norm Spectrum Estimation From Uniformly Spaced Measurements , 2014, IEEE Transactions on Automatic Control.

[24]  Manfred Morari,et al.  System identification via nuclear norm regularization for simulated moving bed processes from incomplete data sets , 2009, Proceedings of the 48h IEEE Conference on Decision and Control (CDC) held jointly with 2009 28th Chinese Control Conference.

[25]  Gregory C. Reinsel,et al.  An Exact Maximum Likelihood Estimation Procedure for Regression- ARMA Time Series Models with Possibly Nonconsecutive Data , 1986 .

[26]  Jan Swevers,et al.  A subspace algorithm for the identification of discrete time frequency domain power spectra , 1997, Autom..

[27]  Alf Isaksson,et al.  Identification of ARX-models subject to missing data , 1993, IEEE Trans. Autom. Control..

[28]  Hüseyin Akçay,et al.  Subspace-based spectrum estimation in frequency-domain by regularized nuclear norm minimization , 2014, Signal Process..