Machine-Learning Models of Components in Electromagnetic Compatibility

Solving electromagnetic problems of complex geometries requires long computation times, preventing tasks where thousands or even millions of simulations runs over ranges of > 5 input parameters are desirable. Examples include multi-objective optimization, risk analysis, uncertainty quantification, and model calibration. Training a machine-learning model based on 3D simulation reduces the effort to a much lower number of 3D simulations. Once the machine-learning model is ready, it computes within some 10 milliseconds, bringing the analysis of 1 million samples comes within the range of a few days of computation time on a standard computer. We present several examples of machine-learning models with 4 to 15 design parameters and their application in electromagnetic compatibility (EMC), such as a common mode choke, a high-voltage filter, and a half bridge.

Presenter: Jan HANSEN, Graz University of Technology

Replay	Slide deck

Enabling Radar Sensor Vehicle Integration by electromagnetic simulations

Automotive radar is a key sensor system enabling object detection and tracking. An unmounted radar has ideal characteristics. However, when Radar is integrated inside the vehicle, vehicle body and other components in the vicinity such as bumper, chassis, cables influence the electromagnetic waves emanating from it and deteriorates its performance. In such a scenario, electromagnetic simulation can be utilized for incorporating design changes in early stages of product development. Due to the very high frequency application around 77GHz, and the large electrical size of the complex structures around the sensor, ideal simulation techniques are required to optimize the radar integration scenarios. Hence, the key challenge here is to arrive at the optimal trade-off between accuracy and computational resources/simulation time. The presentation would outline how radar vehicle integration is enabled from an electromagnetic standpoint and how optimal solutions are derived by leveraging RF simulation using SIMULIA CST Studio Suite. Various radar vehicle integration scenarios and associated RF simulation workflows would be detailed so as to illustrate how the key challenges are tackled.

Presenter: Yadhu KRISHNAN M K, Continental AG

Radar Sensor Integration into Vehicles using 3DEXPERIENCE® Platform

Presenter: Jan EICHLER, Dassault Systèmes

Replay	Slide deck