Jan Hansen

Achieving first-time-right electromagnetic compatibility (EMC) is crucial in automotive industry for competitiveness. However, compliance with EM emission limits is usually evaluated through measurements on hardware prototypes, especially for complex system like electric vehicle (EV) powertrains, which include multiple components such as traction inverters (TIs) that collectively contribute to the overall emissions. Furthermore, several design goals, such as the amount of magnetic material or total installed filter capacitances, must be considered. Therefore, this paper introduces a simulation framework based on common- and differential-mode emission surrogate models with 16 design parameters. The models are derived from a circuit simulation model valid up to $30 \,\mathrm{M}\mathrm{Hz}$ and enable the prediction and optimization of conducted emissions (CEs) of traction inverters in the concept phase, prior to hardware prototyping. A Two-Level Traction Inverter (TLTI) system and CISPR 25 standard limits serve as an example. According to the feature selective validation (FSV) metric, the surrogate models achieve a worst-case accuracy rated as “very good” for CE prediction when trained on a dataset of 8064 samples. This corresponds to a spectral uncertainty of approximately $\pm 2 \,\mathrm{dB}$. With evaluation times of about $15 \,\mathrm{m}\mathrm{s}$, the presented surrogate models allow to perform 100 million TI optimizations in just $48 \,\mathrm{h}$, resulting in three selected optimal designs verified by circuit simulations. Therefore, the framework acts as a potential tool to develop first-time-right EMC and performance optimized prototypes.

Wide bandgap semiconductor devices enable improvements in power electronics efficiency and power density. However, their fast switching characteristics exacerbate high frequency electromagnetic interference challenges. Radiated emission prediction methods typically employ measured or calculated common-mode signals as AC noise sources based on transistor switching waveforms. These approaches are limited in bandwidth and are dependent on operation points. This work presents a novel measurement based modeling methodology with a measurement setup using a Gaussian Process Regression ($\mathcal {GPR}$) model to overcome the limitations by predicting the transistors drain voltage and source current in the frequency domain up to 1 GHz. We develop a setup to acquire the switching signals of a transistor in frequency domain across various operating conditions within a six dimensional input parameter space. Demonstrating in this article the simulation-based proof of concept of this approach, that a $\mathcal {GPR}$ model can predict the source current and the drain voltage of a transistor in frequency domain. The $\mathcal {GPR}$ model, which is trained in this work using simulation data, predicts the transistors source current in the frequency domain for unseen operating points with a maximum NRMSE of $2.1 \,\%$. These results are establishing a novel approach for noise source modeling in radiated emission models of power converter systems.

Zusammenfassung: Elektrische Maschinen innerhalb von E‑Fahrzeugen sind relevante Pfade für die Auskopplung elektromagnetischer Emissionen. Für die Einhaltung der elektromagnetischen Verträglichkeit (EMV) ist deren Modellierung von entscheidender Bedeutung, aufgrund komplexer Geometrien jedoch mit hohem Rechenaufwand verbunden. In diesem Beitrag wird ein Multi-Fidelity-Surrogate-Modellierungsansatz vorgeschlagen, der Low-Fidelity-Schaltungssimulationen mit High-Fidelity-3D-Elektromagnetiksimulationen kombiniert. Es wird eine Kriging-Methode vorgestellt, bei der frequenzbezogene Korrelationen verwendet werden, um LF- und HF-Daten zu kombinieren. Die Methode wird anhand eines vereinfachten Modells einer elektrischen Maschine demonstriert. Im Vergleich zu herkömmlichen MF-Ansätzen verbessert die vorgeschlagene Methode die Genauigkeit bei Resonanzstellen und reduziert gleichzeitig die Anzahl notwendiger 3D-Elektromagnetiksimulationen.

To evaluate the system-level electromagnetic emissions of an automobile, in a simulation environment, individual sources of electromagnetic interference (EMI) must be modeled. Modeling the emission behavior of electronic controller units (ECUs) in automobiles requires the characterization of the main microcontroller as an emission source. The conducted emissions (CEs) of the microcontroller depend on the programs running inside the core and the functional blocks utilized for those programs. We propose a measurement-based stochastic model to characterize such program dependency on the conducted emissions of a microcontroller. From board-level CE measurements and the functional blocks utilized for the programs, we obtain the upper and lower bounds of the emissions within which the microcontroller’s conducted emissions are expected to lie with a 95% confidence interval. Further, we propose a method to estimate the emission peaks of the microcontroller when a new program runs in its core with a ±4 dB average error. The functional blocks used for this analysis involve the RAM, timers, and GPIOs. The method works satisfactorily at frequencies up to 500 MHz and is tested on an STM32 general-purpose microcontroller board.

This article describes the Cell Maps for Artificial Intelligence (CM4AI) project and its goals, methods, standards, current datasets, software tools, status, and future directions. CM4AI is the Functional Genomics Data Generation Project in the U.S. National Institute of Health’s (NIH) Bridge2AI program. Its overarching mission is to produce ethical, AI-ready datasets of cell architecture, inferred from multimodal data collected for human cell lines, to enable transformative biomedical AI research.

Employing multiobjective optimization (MOO) for grid-based 3-D electromagnetic simulations is computationally expensive. Using machine learning models as surrogates can mitigate this problem. However, for complex electromagnetic compatibility (EMC) systems, surrogates become inefficient since the required number of training samples leads to high model generation times. Furthermore, in EMC, the functions are often described in a frequency range. This demands a vectorial approach to the problem, which most methods cannot provide. In this work, we propose a Gaussian process model, called wideband Kriging, to generate a vector-valued model over frequency. We study, as an example, a high-voltage electromagnetic interference (EMI) filter's MOO over a design space of 17 parameters. The wideband Kriging needs only 450 training samples to achieve an accuracy of $<\!\!\text{1.5}\%$ for a frequency range from 1 kHz to 400 MHz. The model runs 800 times/minute, enabling its use for high dimensional MOOs. We find the best filter design concerning volume, current rating, electromagnetic compatibility (EMC) performance, and saturation.

The increasing application of silicon carbide and gallium nitride transistors decreases the size of power electronic systems because of faster switching and clock times. However, both increase the electromagnetic emission, causing additional cost for the design of EMI filters. Finding a joint optimum requires long computation time. Meta modeling techniques reduce the computation time by approximating physical models with a simple mathematical equivalent function, generated from training data. If the physical models are computationally expensive, it is important to identify particularly efficient methods. We compare the surrogate modeling techniques of Kriging, Polynomial Chaos Expansion, Polynomial Chaos Kriging and Wideband Kriging. We find that Wideband Kriging is most accurate, but also has the most excessive model generation time.

In Multi-Objective optimization (MOO) many conflicting key performance indicators (KPIs) require a large number of simulations to be carried out to obtain the Pareto fronts. The surrogate model-based approach allows faster model evaluation, easing the burden of MOO. This document presents the MOO of a Kriging-based metamodel of a half bridge circuit combined with an electromagnetic interference (EMI) filter using non-dominated sorting genetic algorithm II (NSGA-II) to comply with EMC limits. Out of a total of 14 design parameters which concern both the functional and the EMI design of the system, we study five dimensions in detail and obtain Pareto-optimal designs with a good compromise between fast switching and EMC compliance.