Abstract: Given measurements of a linear time-invariant system, the McMillan degree is the dimension of the smallest such system that reproduces these observed dynamics. Using impulse response measurements where the system has been started in some (unknown) state and then allowed to evolve freely, a classical result by Ho and Kalman reveals the McMillan degree as the rank of a Hankel matrix built from these measurements. However, if measurements are contaminated by noise, this Hankel matrix will almost surely be full rank. Hence practitioners often estimate the rank of this matrix---and thus the McMillan degree---by manually setting a threshold between the large singular values that correspond to the non-zero singular values of the noise-free Hankel matrix and the small singular values that are pertubations of the zero singular values. Here we introduce a probabilistic upper bound on the perturbation of the singular values of this Hankel matrix when measurements are corrupted by additive Gaussian noise, and hence provide guidance on setting the threshold to obtain a lower bound on the McMillan degree. This result is powered by a new, probabilistic bound on the 2-norm of a random Hankel matrix with normally distributed entries. Unlike existing results for random Hankel matrices, this bound features no unknown constants and, moreover, is within a small factor of the empirically observed bound when entries are independent and identically distributed. This bound on the McMillan degree provides an inexpensive alternative to more general model order selection techniques such as the Akaike Information Criteria (AIC).
Recommendations
- scientific article; zbMATH DE number 917713
- Structure and order estimation of multivariable stochastic processes
- All optimal Hankel-norm approximations of linear multivariable systems and theirL,∞-error bounds†
- Model order determination using the Hankel matrix of impulse responses
- Stochastic model reduction via Hankel-approximation
Cites work
- \({\mathcal H}_2\)-optimal model reduction with higher-order poles
- A few remarks on the operator norm of random Toeplitz matrices
- A new look at the statistical model identification
- An eigensystem realization algorithm for modal parameter identification and model reduction
- ARPACK Users' Guide
- Dynamic system identification. Experiment design and data analysis
- scientific article; zbMATH DE number 1347881 (Why is no real title available?)
- scientific article; zbMATH DE number 1182386 (Why is no real title available?)
- scientific article; zbMATH DE number 3997615 (Why is no real title available?)
- scientific article; zbMATH DE number 2118874 (Why is no real title available?)
- scientific article; zbMATH DE number 3233996 (Why is no real title available?)
- scientific article; zbMATH DE number 2222868 (Why is no real title available?)
- Matrix Analysis
- Minimal state-space realization in linear system theory: An overview
- Model Selection and Multimodel Inference
- On the finite rank and finite-dimensional representation of bounded semi-infinite Hankel operators
- On the spectral norm of a random Toeplitz matrix
- Remark on the norm of random Hankel matrices
- Spectral measure of large random Hankel, Markov and Toeplitz matrices
- The fit of a sum of exponentials to noisy data
Cited in
(2)
This page was built for publication: A data-driven McMillan degree lower bound
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5146675)
