12,000

We have over 12,000 students, from over 100 countries, within one of the safest campuses in the UK

93%

93% of Lancaster students go into work or further study within six months of graduating

Home > Research > Publications & Outputs > Two approaches to data-driven design of evolvin...
View graph of relations

« Back

Two approaches to data-driven design of evolving fuzzy systems: eTS and FLEXFIS

Research output: Contribution in Book/Report/ProceedingsPaper

Published

Publication date22/06/2005
Host publicationFuzzy Information Processing Society, 2005. NAFIPS 2005. Annual Meeting of the North American
PublisherIEEE
Pages31-35
Number of pages5
ISBN (Print)0-7803-9187-X
Original languageEnglish

Conference

Conference2005 Annual Meeting of the North American Fuzzy Information Processing Society Annual Conference
CityAnn Arbor, Michigan, USA
Period21/06/0525/06/05

Conference

Conference2005 Annual Meeting of the North American Fuzzy Information Processing Society Annual Conference
CityAnn Arbor, Michigan, USA
Period21/06/0525/06/05

Abstract

In this paper two approaches for the incremental data-driven learning of one of the most effective fuzzy model, namely of so-called Takagi-Sugeno type, are compared. The algorithms that realise these approaches include not only adaptation of linear parameters in fuzzy systems appearing in the rule consequents, but also incremental learning and evolution of premise parameters appearing in the membership functions (i.e. fuzzy sets) in sample mode together with a rule learning strategy. In this sense the proposed methods are applicable for fast model training tasks in various industrial processes, whenever there is a demand of online system identification in order to apply models representing nonlinear system behaviors to system monitoring, online fault detection or open-loop control. An evaluation of the incremental learning algorithms are included at the end of the paper, where a comparison between conventional batch modelling methods for fuzzy systems and the incremental learning methods demonstrated in this paper is made with respect to model qualities and computation time. This evaluation is based on high dimensional data coming from an industrial measuring process as well as from a known source on the Internet, which underlines the usage of the new method for fast online identification tasks.