Usporedba rješenja višeciljne optimizacije na primjeru dimenzioniranja jednostavne brodske konstrukcije

master's thesis

• Title (english): Comparison of multi-objective optimization solutions on the example of designing a simple ship structure
• Author: Veron Hrvojić
• Mentor: Pero Prebeg (mentor)
• Committee member: Tomislav Stipančić (predsjednik povjerenstva)
• Committee member: Pero Prebeg (član povjerenstva)
• Committee member: Matija Hoić (član povjerenstva)
• Committee member: Milan Kostelac (član povjerenstva)
• Committee member: Goran Šagi (član povjerenstva)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2022-07-20, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Shipbuilding; Constructive of Floating and Maritime Structures

Abstract

U diplomskom radu provedena je usporedba kvalitete rezultata optimizacije višeciljnog problema s ograničenjima ostvarenih različitim optimizacijskim algoritmima iz dviju Python biblioteka. Usporedba je, osim na jednostavnom standardnom problemu, provedena i na primjeru okvira jednostavne brodske konstrukcije s tri palube i tri tanka ispod najdonje palube, pri čemu je za analizu odziva korištena metoda pomaka. Da bi se omogućilo formuliranje optimizacijskog problema neovisno o biblioteci koja se koristi za optimizaciju te za automatizirano provođenje optimizacija više optimizacijskih problema optimizacije i njezinu evaluaciju, nadograđena je biblioteka za usporedbu rješenja višeciljnih optimizacijskih problema moobench. Biblioteka moobench softver je otvorenog koda dostupan na mrežnom servisu GitHub. U okviru rada implementirana su sučelja prema Python bibliotekama Pymoo i jMetalPy i dodana toj biblioteci čime su optimizacijski algoritmi iz tih dvaju biblioteka postali dostupni biblioteci moobench. Isto tako, u biblioteku je implementiran i optimizacijski problem dimenzioniranja okvira jednostavne brodske konstrukcije. Usporedba kvalitete rješenja temeljena je na mjerama kvalitete Pareto fronte koje su u radu također implementirane u moobench.

Abstract (english)

In final thesis a comparison on multiobjective constrained optimization results was conducted with a variety of optimization algorithms from two Python libraries. A comparison is done, on a standard benchmark problem, and on model of a simple ship structure, with three decks and three tanks below the lowest deck, optimization problem. The displacement method is used to determine simple ship structure frame response, which is also implemented in Python. In order to enable optimization problem formulation independent of library used for optimization and to automate optimization procedure for multiple runs and evaluation of results, a framework for comparison of multiobjective optimization results moobench was used and extended. Moobench is open-source software available at GitHub. In scope of this thesis, two interfaces to Python libraries Pymoo and jMetalPy were implemented and added to moobench, so that optimization algorithms from those two libraries are now available in moobench. Also, a model of simple ship frame design optimization problem is added to moobench. Quality comparison of solutions is based on perfomance indicators for Pareto fronts that are also implemented in moobench in a scope of this thesis

• Keywords: moobench; optimizacija; usporedba; mjere kvalitete; Pareto
• Keywords (english): moobench; optimization; comparison; qualitiy indicators; performance indicators; Pareto
• Language: croatian
• URN:NBN: urn:nbn:hr:235:831099
• Study programme: Title: Mechanical Engineering; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka strojarstva (magistar/magistra inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2022-07-18 14:33:48