Machine Learning for Grey Box Modelling of Electrical Components for Circuit- and EMC-Simulation

Konferenz: PCIM Europe digital days 2020 - International Exhibition and Conference for Power Electronics, Intelligent Motion, Renewable Energy and Energy Management
07.07.2020 - 08.07.2020 in Deutschland

Tagungsband: PCIM Europe digital days 2020

Seiten: 9Sprache: EnglischTyp: PDF

Autoren:
Roche, Jan-Philipp (KEB Automation KG, Germany)
Friebe, Jens (Institute for Drive Systems and Power Electronics (IAL), Leibniz University Hannover, Germany)
Niggemann, Oliver (Helmut Schmidt University/University of the Federal Armed Forces Hamburg, Germany)

Inhalt:
Models of electrical components are needed for the simulation of electrical circuits. Typically a function based on currents and voltages is the main result of these simulations. Based on the currents and voltages, the EMC can be evaluated and improved systematically. Unfortunately, sufficient simulation models are often not available, especially when it comes to EMC behavior. Grey box learning is a promising approach to fill this gap. The development of a modelling concept is the focus of this paper.