• TUCS Publication Series
• How to Publish in TUCS Series
• Publication Search
• Publication Graph
• Publication Input
• Publication Input Guide
• JuFo Browser

Least-Squares Support Vector Regression for Identification of Periodic Systems

Lassi Hietarinta, Hannu T. Toivonen, Least-Squares Support Vector Regression for Identification of Periodic Systems. In: Proceedings IFAC Workshop on Periodic Control Systems (PSYCO 2010), 2010.

Abstract:

Least-squares support vector regression is applied to identify linear periodically time-varying (LPTV) systems. In this contribution the LPTV systems are modelled using a Fourier series representation of the transfer function and orthonormal filter expansions, such as Laguerre or Kautz filters. The model parameters are determined using least-squares support vector regression. In order to decrease complexity of the identified model, smoothness priors which penalize fluctuations of the frequency response can be introduced in the support vector procedure. The performance of the identified models are assessed by means of several simulation trials.

BibTeX entry:

@INPROCEEDINGS{inpHiTo10b,
  title = {Least-Squares Support Vector Regression for Identification of Periodic Systems},
  booktitle = {Proceedings IFAC Workshop on Periodic Control Systems (PSYCO 2010)},
  author = {Hietarinta, Lassi and Toivonen, Hannu T.},
  year = {2010},
}

Belongs to TUCS Research Unit(s): Other

Edit publication