Adaptivno vodenje osnovano na algoritmih računske inteligence

magistrsko delo

Jakob Šafarič (Author), Iztok Fister (Mentor), Darko Lovrec (Mentor), Božidar Bratina (Co-mentor)

Abstract

Na področju robotike obstaja ogromno nelinearnih sistemov, ki se še vedno vodijo z linearnimi regulatorji, čeprav ti niso optimalna rešitev za dani problem. V tem magistrskem delu je predstavljen hitrostni adaptivni nelinearni regulator, ki je sposoben voditi nelinearno progo boljše kot linearni regulatorji. Razviti regulator je sestavljen iz algoritma po vzorih iz narave, ki optimira vrednost referenčnega toka, in umetne nevronske mreže, ki je sposobna napovedati vrednost ocenitvene funkcije za izbrani algoritem. Pri tem primerjamo vpliv različnih algoritmov po vzorih iz narave na delovanje predlaganega regulatorja. V naši primerjalni analizi smo zajeli naslednje algoritme: evolucijsko strategijo, diferencialno evolucijo, optimizacijo z roji delcev in algoritmom po vzoru obnašanja netopirjev.

Keywords

nelinearni adaptivni regulatorji;umetne nevronske mreže;evolucijski algoritmi;algoritmi inteligence rojev;magistrske naloge;

Data

Language:	Slovenian
Year of publishing:	2018
Typology:	2.09 - Master's Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	J. Šafarič
UDC:	004.434:004.8(043.2)
COBISS:	21806102
Views:	1118
Downloads:	249
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Adaptive control based on computational intelligence
Secondary abstract:	In robotics, there are a lot of nonlinear systems, which are still controlled using linear controllers, even though they are not optimal solutions for the given problem. In this work, an adaptive nonlinear velocity controller is presented, which is better suited for control of nonlinear systems. The presented controller consists from nature inspired algorithm, which optimizes current reference, and artificial neural network, which is used for fitness function evaluation. A comparison of controller operation, when different nature inspired algorithms are used, is also presented in this work. Thus, the following nature inspired algorithms were captured in our comparison study: evolutionary strategy, differential evolution, particle swarm algorithm and bat algorithm.
Secondary keywords:	nonlinear adaptive controllers;artificial neural networks;evolution algorithms;swarm intelligence algorithms;
URN:	URN:SI:UM:
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Mehatronika
Pages:	X, 61 str.
ID:	10955078