Physics guided model for learning and control for EMC
Host: TUe (NL)
Main supervisor: Dr. Amritam Das (TUe, NL)
Co-supervisors/mentors: Prof.dr.ir. Roland Toth (TUe, NL), Prof.dr.ir Tim Claeys (KUL, BE), Ir. Rob Kleihorst (Philips, NL)
Duration: 36 months
Required profile: Electrical Engineering Desirable skills/interests: Machine Learning, Data-driven modelling, Control design, Electromagnetism
Objectives: Modelling EMC is challenging due to nonlinear behavior, frequency-dependent interactions, complex geometries and many uncertainties involved. Current technqiues are heavily limited due computational burden and are often too costly to enable optimization early in the design process. This DC will aim to capture the time-varying behaviour of components, connectivity, and operational conditions with machine learning methods, such as Physics-guided Neural Networks (PGNNs) and explore control strategies to design controllers efficiently based on the learnt models to suppress electromagnetic interference within electronic systems. The DC will also support capturing time-varying behavior of EMC and contribute to the desing of a medical collaborative system.
Secondments (2-4 months in total): KU Leuven (BE), Philips (NL)
Apply: https://pattern-dn.eu/index.php/esr-projects/
AI to combine and model Electromagnetic Noise Footprint (EMNF) applied to cables.
Host: TUe (NL)
Main supervisor: Dr. Ir. Anne Roc’h
Co-supervisors/mentors: Prof. Philippe Besnier (IETR (CNRS), FR)/ Ir. Rob Kleihorst (Philips, NL)/ Prof.dr.ir. Roland Toth (TUe, NL)
Duration: 36 months
Required profile: Electronic Engineering Desirable skills/interests: Electromagnetism, Electromagnetic Compatibility, Metrology, Artificial Intelligence.
Objectives: This DC first goal will be to unravel a structural understanding of the exchange of parasitic energy of cables with its environment using AI tools. The so-called EMNF (ElectroMagnetic Noise Footprint) comprises a set of characteristic curves obtained from stand-alone measurements on a device. A second goal will consist in exploring how to combine two or more EMNFs. The work will support an optimization (within the SSbD framework) of cable routing in a MedTech product by combining EMNFs.
Secondments (2-4 months in total): Safran (FR) and IETR (CNRS) (FR).
Apply: https://pattern-dn.eu/index.php/esr-projects/
AI to combine and model Electromagnetic Noise Footprint (EMNF) applied to PCB Tracks
Host: TUe (NL)
Main supervisor: Dr.ir Anne Roc’h (TUe, NL)
Co-supervisors/mentors: Prof. Davy Pissoort (KUL, BE)/ David Kuratko (IDIADA, CZ)/ Prof.dr.ir. Roland Toth (TUe, NL).
Duration: 36 months
Required profile: Electrical Engineering
Desirable skills/interests: Electromagnetism, Electromagnetic Compatibility, Metrology, Artificial Intelligence.
Objectives: This DC first goal will be to unravel a structural understanding of the exchange of parasitic energy a PCB Tracks with its environment using AI tools. The so-called EMNF (ElectroMagnetic Noise Footprint) comprises a set of characteristic curves obtained from stand-alone measurements on a PCB. A second goal will consist in exploring how to combine two or more EMNFs. The work will support an optimization (within the SSbD framework) of PCB routing by combining EMNFs.
Secondments (2- 4 months in total): KU Leuven (BE) and Idiada (CZ).