Machine Learning Barycenter Approach to Identifying LPV State-Space Models

• Type: Article in International Conference Proceedings Book
• Date: 2016

Authors

Romano, RA (Author) Other; P. Lopes dos Santos (Author) FEUP; Pait, F (Author) Other; Perdicoulis, TP (Author) Other; Ramos, JA (Author) Other

Conference proceedings International

Title: 2016 AMERICAN CONTROL CONFERENCE (ACC)

Pages: 6351-6356

American Control Conference (ACC)

Boston, MA, JUL 06-08, 2016

Indexing

ISI Web of Knowledge - 6 Citations

Scopus - 9 Citations

Other information

Authenticus ID: P-00M-HJB

DOI: 10.1109/acc.2016.7526668

Abstract (EN): In this paper an identification method for statespace LPV models is presented. The method is based on a particular parameterization that can be written in linear regression form and enables model estimation to be handled using Least-Squares Support Vector Machine (LS-SVM). The regression form has a set of design variables that act as filter poles to the underlying basis functions. In order to preserve the meaning of the Kernel functions (crucial in the LS-SVM context), these are filtered by a 2D-system with the predictor dynamics. A data-driven, direct optimization based approach for tuning this filter is proposed. The method is assessed using a simulated example and the results obtained are twofold. First, in spite of the difficult nonlinearities involved, the nonparametric algorithm was able to learn the underlying dependencies on the scheduling signal. Second, a significant improvement in the performance of the proposed method is registered, if compared with the one achieved by placing the predictor poles at the origin of the complex plane, which is equivalent to considering an estimator based on an LPV auto-regressive structure.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 6