Affinely parametrized state-space models

Affinely parametrized state-space models: Ways to maximize the Likelihood Function

Wills, Adrian (University of Newcastle, Australia)
Yu, C. (Beijing Institute of Technology)
Ljung, Lennart (Linköping University)
Verhaegen, M.H.G. (TU Delft Team Raf Van de Plas)

2018

Using Maximum Likelihood (or Prediction Error) methods to identify linear state space model is a prime technique. The likelihood function is a nonconvex function and care must be exercised in the numerical maximization. Here the focus will be on affine parameterizations which allow some special techniques and algorithms. Three approaches to formulate and perform the maximization are described in this contribution: (1) The standard and well known Gauss-Newton iterative search, (2) a scheme based on the EM (expectation-maximization) technique, which becomes especially simple in the affine parameterization case, and (3) a new approach based on lifting the problem to a higher dimension in the parameter space and introducing rank constraints.

difference-of-convex optimization
expectation-maximization algorithm
maximum-likelihood estimation
Parameterized state-space model

http://resolver.tudelft.nl/uuid:18d98999-485e-409e-8a37-e1c50a9dfcf1

https://doi.org/10.1016/j.ifacol.2018.09.170

2405-8963

IFAC-PapersOnLine, 51 (15), 718-723

SYSID 2018: 18th IFAC Symposium on System Identification, 2018-07-09 → 2018-07-11, Stockholm, Sweden

Institutional Repository

journal article

© 2018 Adrian Wills, C. Yu, Lennart Ljung, M.H.G. Verhaegen