31. Annual Meeting (2024) 27.11.2024, 11 Uhr In-person: Helmut-Schmidt-Universität – Universität der Bundeswehr Hamburg Via Zoom: Meeting-URL: https://jade-hs.zoom-x.de/j/63490796275?pwd=yP6xQubs4BT4HGObeXQX2YLOS82w8v.1 Meeting-ID: 634 9079 6275 Kenncode: 995643
EMC Professional Talk: EMC on HV system level for electric vehicles DI Dr. techn. Guido Rasek Valeo eAutomotive, Erlangen, Germany 19.11.2024, 4:00 p.m. (UTC+1) Abstract: Noise currents in high-voltage power trains of electric vehicles place significant strain on the EMC filters used to mitigate conducted and radiated emissions. Filtering targets for high levels emissions must be achieved. The power dissipation caused by the resistive properties of the filter components is a key factor in determining the temperature behavior and lifetime of the filter. A validated numerical EMC model of an electric vehicle power train can be used to complete investigations. Principle outlines of the activities to achieve postulated goals are presented. For further information, pleas see event page: EMC Professional Talk: EMC on HV system level for electric vehicles
EMC Professional Talk: PIML-Driven EMI/EMC Simulations Prof. Mohamed Kheir University of Southern Denmark, Sønderborg, Denmark Abstract: Electromagnetic Interference and Compatibility (EMI/EMC) are crucial aspects of electrical and electronic devices. There are many standards that strictly regulate the emission levels allowed from any electrical device. EMI simulation tools play an essential role during product design by investigating any unintentional emissions before compliance testing. However, these tools suffer from several problems that make them complicated and not environmentally friendly. For instance: 1) long simulation time, 2) computational complexity, and 3) high energy consumption. All these issues make EMI simulations non-green and costly. One potential solution to these problems is utilizing Machine Learning (ML) as a green alternative to traditional simulations. The fusion of Physics-Informed Machine Learning (PIML) with conventional ML techniques has emerged as a transformative force, seamlessly integrating domain-specific physics with ML. For example, in an EMC setting, it can encode Maxwell’s equations and fundamental principles, bridging theory and data. This proposed approach can reduce dependence on computationally expensive simulations and accelerate EMI modeling without the need for structure discretization as in traditional numerical simulations. For further information, pleas see event page: EMC Professional Talk-PIML-Driven EMI/EMC Simulations
