Odjel za industrijske primjene Hrvatske sekcije IEEE i Zavod za elektrostrojarstvo i automatizaciju FER-a pozivaju vas na online predavanje:
"Meta-model-based sensitivity analysis and optimisation for electromagnetic design tasks"
koje će održati Dr. Ing. Markus Stokmaier, Dynardo GmbH, Weimar, Njemačka. Predavanje će se održati u četvrtak, 26. studenog 2020. godine s početkom u 11:00 sati.
Poveznica na predavanje nalazi se ovdje.
Životopis predavača i sažetak predavanja nalaze se u nastavku obavijesti.
Životopis predavača
Dr. Ing. Markus Stokmaier graduated with a Diploma in Physics at the University of Karlsruhe in 2006. In 2002-2003 he studied at the Université Joseph Fourier in Grenoble where he received the Maîtrise de Physique degree. He received Dr.Ing. degree from Institute for Nuclear and Energy Technologies (IKET) at Karlsruhe Institute of Technology (KIT) in 2020. From 2006 to 2015 he worked at the Karlsruhe Institute of Technology's Institute of Nuclear and Energy Technologies and at Rensselaer Polytechnic Institute (Troy, NY). Since 2015 he works on workflow and software development at Dynardo GmbH which is under the umbrella of ANSYS Germany since November 2019.
Sažetak predavanja
Simulation-based design of electromagnetic (EM) systems like actuators, motors, antennas etc. offers the potential of systematically searching and finding optimal designs. When seeking innovative solutions to technical challenges, there is always a tension between intuition and formalisation. How much effort should be invested into formalised parametrisation, formalised objectives and constraints, and automated setups for design space exploration with algorithms? How much of that is beneficial without quenching creativity based on free thinking and intuition?
In the first part, this presentation outlines the basics of methods like generic meta-model regression, meta-model-based sensitivity analysis, and-box optimisation algorithms. Two focus points are the cross-validation-based Coefficient of Prognosis (CoP) and global optimisation with evolutionary algorithms (EA). The second part leaves the black-box perspective behind and discusses case studies of electric machine and antenna design. They describe exemplary approaches of dealing with goal conflicts arising e.g. from the multi-physical nature of electric machines. The case studies aim at underlining that generic techniques for meta-modelling and algorithmic optimisation are best used when they strengthen at the same physical system understanding, engineering intuition, and quantitative treatment.
