IDENTIFICATION OF NONLINEAR SYSTEMS USING FIXED BUDGET KERNEL LMS ALGORITHM
Journal Title: Informatyka Automatyka Pomiary w Gospodarce i Ochronie Środowiska - Year 2012, Vol 2, Issue 4
Abstract
In this paper a new version of kernel normalized least mean squares algorithm is applied to identification of nonlinear system. To maintain a fixed amount of support vectors, requisite for practical kernel-based algorithm, a pruning criterion is used. After admitting a new input vector to the dictionary, a least important entry is selected and discarder. A case of nonlinear system identification is presented, proving that algorithm performs well and it can maintain a performance comparable to state-of-the-art algorithms, using smaller number of support vectors.
Authors and Affiliations
Dominik Rzepka, Piotr Otfinowski
ALGORITHMIC REALIZATION IMPACTS ASSESSMENT FOR RENEWABLE SOURCES ELECTRIC POWER ON RELIABILITY ELECTRICAL NETWORKS
For the effective operation of distributed energy sources and their use in complex electrical power networks, studies their properties, resulting from joint work in the electric power system. Algorithm of performance ass...
BIAŁYSTOK UNIVERSITY OF TECHNOLOGY, FACULTY OF ELECTRICAL ENGINEERING- DEVELOPMENT AND NEW CHALLENGES
The paper gives a glance on history and current educational offer of the Faculty of Electrical Engineering of Białystok University of Technology. Some leading areas of electrical technologies worked out on the faculty a...
ZASTOSOWANIE BIBLIOTEK NUMERYCZNYCH W OBLICZENIACH MEB
Zastosowanie bibliotek numerycznych pozwala na znaczne skrócenie czasu obliczeń i ułatwienie pisania kodu programu. Popularne biblioteki BLAS i LAPACK doczekały się dojrzałych implementacji pozwalających na wykorzystanie...
APPLYING A Q-GRAM BASED MULTIPLE STRING MATCHING ALGORITHM FOR APPROXIMATE MATCHING
We consider the application of multiple pattern matching (Multi AOSO on q-Grams) algorithm for approximate pattern matching. We propose the on-line approach which translates the problem from approximate pattern matching...
REDUKCJA WYMIAROWOŚCI DANYCH A SELEKCJA CECH W ZASTOSOWANIU DO PROGNOZOWANIA MAKSYMALNEGO DOBOWEGO OBCIĄŻENIA ELEKTROENERGETYCZNEGO
Prognozowanie obciążeń w systemie elektroenergetycznym jest ważnym problemem praktycznym zarówno z technicznego jak i ekonomicznego punktu widzenia. W małych systemach problem ten jest stosunkowo trudny do rozwiązania ze...