Naslov
Neural Networks and Wavelets in Ships and Autopilots Identification

Autori
Vujović, Igor ; Beroš, Slobodan ; Kuzmanić, Ivica

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Proc. of 14th International Scientific and Professional Congress Theory and Practice of Shipbuilding / Čalić, Bruno - Rijeka : Faculty of Engeenering, University of Rijeka, 2000, 149-156

ISBN
953-6320-29-8

Skup
14th International Scientific and Professional Congress Theory and Practice of Shipbuilding

Mjesto i datum
Rijeka, Hrvatska, 23.11.2000. - 25.11.2000

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
wavelet; neural network; system identification; autopilot

Sažetak
A referent model is used in autopilot design nowadays, where the parameters of the model are identified. Such model is not adjusted for a single particular ship. Introduction of self-learning neural networks enables individual approach to every ship and design of the best possible control for the particular ship. The basic mathematic and algorithms are discussed. Two types of neural networks are considered: feed-foreward networks (where neurons are organised in layers and typical application is in the non-linear static mapping) and recurrent network (where the output of neurons are fed back to the inputs of some neurons. These networks are used in non-linear dynamic systems). Wavelets' role in ships' identification is considered, as well as the wavelet networks' role, as the type of the neural networks. This article presents a general idea of wavelet and neural network application in ship's identification and modelling, as well as the possible usage of wavelets aboard ships.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika

POVEZANOST RADA