Višeciljno optimiranje električnih vučnih motora temeljeno na metamodeliranju

master's thesis

• Title (english): Multi-objective optimization of electrical traction motors based on metamodeling
• Author: Lucija Miličević
• Mentor: Stjepan Stipetić (mentor)
• Committee member: Stjepan Stipetić (predsjednik povjerenstva)
• Committee member: Damir Žarko (član povjerenstva)
• Committee member: Marinko Kovačić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-02-27, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Prilikom projektiranja električnih motora potrebno je voditi računa o toplinskim, mehaničkim i elektromagnetskim aspektima motora. Pronalazak dizajna koji će zadovoljiti unaprijed definirane karakteristike rada često uključuje više oprečnih ciljeva, primjerice, učinkovitost, troškovi, težina motora. Svladavanje takvog problema omogućuje višeciljna optimizacija. Predložen je novi optimizacijski pristup koji se temelji na metamodeliranju (Metamodel of Optimal Prognosis - MOP). CoP (Coefficient of Prognosis) je mjera kvalitete aproksimacije koja opisuje metamodeliranje. Pomoću MOP metode pronalazi se najbolji n-dimenzionalni prostor ulaznih parametara, najbolji metamodel za svaku izlaznu veličinu te procjenjuje kvalitetu aproksimacije. MOP metoda započinje provođenjem analize osjetljivosti nad svim dizajnima generiranim pomoću unaprijed definirane funkcije za uzorkovanje (Space filling LHS). Svaki dizajn se podvrgava proračunu u alatu Motor-CAD da bi se dobile izlazne performanse pojedinog motora. OptiSLang, pomoću Python i MATLAB skripti, upravlja Motor-CADom kako bi dodijelio parametre, pokrenuo proračune i dohvatio sve relevantne podatke u optiSLang. Nakon analize osjetljivosti generiraju se metamodeli. Metamodeli predstavljaju matematičke funkcije koje prikazuju performanse motora u ovisnosti o svim ulaznim parametrima. Zatim, ako je vrijednost CoP koeficijenta veća ili jednaka $90\%$ moguće je direktno na generiranim metamodelima primijeniti višeciljnu optimizaciju za zadana ograničenja i ciljeve. Kako bi se ispitala učinkovitost gornje metode, provesti će se analiza osjetljivosti i višeciljna optimizacija temeljena na MOP pristupu na tri sinkrona motora s permanentnim magnetima koji imaju različite topologije rotora.

Abstract (english)

Electric motor can only be properly designed considering thermal, mechanical and electromagnetic aspects. Finding the optimal design which will fulfill required performances usually involves multiple criteria that can be in conflict for an instance, efficiency, costs, machine weight. Multi-objective optimization is performed to overcome multi-physics problem. New optimization approach based on Metamodel of Optimal Prognosis (MOP) is proposed. This approach is described by the Coefficient of Prognosis (CoP) which is the measure of the approximation quality. The MOP method finds the best input design space, best metamodel for each response and estimates the approximation quality. The approach starts with sensitivity analysis on the electric motor designs generated by the Space filling Latin Hypercube Sampling function. Each motor sample is evaluated in Motor-CAD in order to generate performance data sets, for an instance torque, power, efficiency. Motor-CAD is driven by optiSLang using customized Python and MATLAB scripts in order to assign parameters, run calculations and extract all relevant data of interest to optiSLang. After analysis, metamodels are built. Metamodels are basically the best response surfaces that capture the physics of the motor and show how the design space and the output performance correlate with each other. Furthermore, if the CoP values are higher or equal to 90\%, an efficient optimization can be run directly on metamodels for given goals and constraints. In order to test the effectiveness of the above procedure, sensitivity analysis and multi-objective optimization based on MOP will be performed on three IPM motors with different rotor topologies.

• Keywords: višeciljna optimizacija; Space filling LHS funkcija za uzorkovanje; Motor-CAD; optiSLang; metamodeliranje; koeficijent kvalitete aproksimacije
• Keywords (english): Multi-objective optimization; Space filling LHS; Motor-CAD; optiSLang; Metamodel of Optimal Prognosis; Coefficient of Prognosis
• Language: english
• URN:NBN: urn:nbn:hr:168:855950
• Study programme: Title: Electrical Engineering and Information Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 09:17:31