### Active Learning for Optimization of EMC processes

With the ever increasing operating frequencies and powers, EMC has now become a major consideration on any project involving the design, construction, manufacture and installation of electrical and electronic equipment and systems. An important step in the design of components and prediction of EMC related problems is the modeling and simulation. The complexity and performance of electrical and electronic devices as well as the number and range of variables in the design spaces means that many of the Physics-Based (PB) used tools are either too slow or too inaccurate for effective design and optimization. Recently, machine learning (ML) tools and techniques have been increasingly used in the EMC domain either to improve PB approaches or to replace them. In ML, computers are used to probe vast amounts of data for structure. One requirement for the effective ML model building is the availability of these large datasets for the training and testing processes. The generation of simulation data of EMC systems using PB tools is more than often quite expensive and very time consuming.

The main goal of this project is the adaptation and extension of active learning schemes such as Bayesian Optimization (BO) to the realm of EMC. In active learning, the expensive to evaluate samples used for training and prediction are intelligently collected at each iteration to reduce the needed number of simulation runs. The active learning methods can be used for optimization tasks, modeling of systems or both at the same time. The main working points can be summarized as:

- Adaptation and foundation-laying of BO-based optimization and model building in the fields of EMC, SI, PI and bio-EMC.
- Establishment of a framework to evaluate the certainty in the solutions provided by the active learning scheme.
- Contribution to the available databases with the generated datasets for the service of the electrical engineering and EMC community.

Bayesian active learning and optimization scheme used to intelligently select the next samples to be simulated and added to the available data. The data is fed to the Bayesian model that predicts the optimal parameters while actively building an accurate model. Source: TET, TUHH

Animation of Bayesian active learning to approximate and maximize a 1D function f(x). The utility function is used at each iteration to pick the next best sample x to observe. The observations are used to create a more accurate approximation of the function and find the maxima. Source: TET, TUHH

**Funding:** Scholar of the Konrad Adenauer Foundation

**Contact: **Youcef Hassab

**Start date:** 02.11.2022

### Analysis of Orbital Angular Momentum (OAM) antennas in Complex Environments

Electromagnetic waves can have a form of moment called the Orbital Angular Momentum (OAM). In waves with this property, the area perpendicular to the direction of propagation does not have a constant phase, as is the case with plane waves. Rather, the phase varies around the center of the wave and this phase distribution rotates in the direction of propagation. From the possible phase distributions and the direction of rotation different orthogonal modes result. This means that the modes do not influence each other and can therefore be used well for communication purposes.

In the past some properties of communication with OAM antennas have been investigated. The properties of OAM antennas shall now be further analyzed from the point of view of EMC.

Dipolar array of eight elements with which the first OAM mode is excited by operating the individual dipoles with a phase difference of ∆Φ=45°. The operating frequency is f=100 MHz. Source: TET, TUHH

Phase of the z-component of the E-field on the surface perpendicular to the direction of propagation. The phase was evaluated in three different distances to the antenna. As antenna a dipole array of eight elements was used, which excites the first OAM mode. The circular sections show how the phase rotates exactly 360° in the azimuthal direction. You can also see how this phase distribution rotates in the direction of propagation. Source: TET, TUHH

**Funding:** DFG

**Contact: **Michael Wulff, M Sc.

**Start date:** 01.12.2020

### EMC of Complex Systems

**Ph. D. Thesis Fabian Happ. 01.04.2012 – 01.09.2017**

In this thesis a numerical method based on the Method of Moments and analytical formulations are developed to calculate electromagnetic fields in the vicinity of Carbon Fiber Composite (CFC) materials. The methods are applied to predict the shielding effectiveness of different CFC materials, to calculate lightning transfer functions for the coupling into an aircraft geometry made of CFC, and to estimate the influence of CFC on the crosstalk between transmission lines.

Apertures in a Computer casing (Source: TET, TUHH).

Current distribution on a field illuminated Computer casing (Source: TET, TUHH).

**Related Publications:**

### HIRF Protection Using Energy Selective Diode Arrays

**Cheng Yang, M. Sc., (China Scholarship Council (CSC)), 01.04.2015 – 17.10.2016**

With more and more electromagnetic energy being transmitted by radar, radio, TV, mobile, hotspot (Wi-Fi), satellite, and other transmitters, detrimental effects of electromagnetic waves on the environment, e.g. disturbance or even damage of electronic equipment, become more and more a concern. To ensure safe operation of various electronic systems in transmitting or receiving states, electromagnetic protection should be considered especially against high intensity radiation fields (HIRFs). This project focuses on diode arrays for implementation of self-activated and energy selective blocking of electromagnetic wave.

Based on the nonlinear behavior of diodes, diode arrays can be shown to be transparent with respect to small field amplitudes and opaque to large field amplitudes. In theory, we expect the structure to work like a spatial switch, while in practice its protection performance is limited by several factors, such as the response time and operating frequency range of the diodes, the transmission loss of the protective structure and near field coupling effects. Therefore, accurate full wave simulation are being used for evaluation of the electromagnetic performance of protective diode arrays. Up to now, mostly infinitely periodic diode arrays have been studied. Next steps include the simulation of finite and non-periodic diode arrays using the electromagnetic field simulator CONCEPT-II.

Geometrical model of diode array designed for vertical polarized protection against HIRF. The diode array consists of conductive strips and miniature diodes on a low dielectric loss substrate of thickness 0.8mm. The horizontal and vertical distance of adjacent diodes is 3.5 mm and 2.25, respectively (figure adapted from “A Novel Method of Energy Selective Surface for Adaptive HPM/EMP Protection,” IEEE APWL, vol. 12, pp. 112–115, 2013.)

Evaluation model of diode array in CONCEPT-II. The array was placed in the front window of a PEC box, and a receiving antenna is positioned inside the PEC box (side length approx. 50 cm). What can be seen are the surface currents on the PEC box (red color indicates high current amplitudes).

**Related Publications: **

### Analysis of Electromagnetic Interference in Server Casings

**Ph.D. Thesis Alexander Vogt. 01.04.2011 – 08.04.2016**

With increasing data rates EM emission has become a major issue for EMC and EMI considerations in current server designs. Major sources of EM emission inside include ICs, cables and connectors, and Printed Circuit Boards. While efficient techniques exist to model the interior coupling of PCB cavities (Zpp ), the external (3D) coupling still has to be computed using full-wave techniques.

This project aims to combine the different methods to model and simulate the components of the server. The dimensions and high data rates in addition to the high complexity of the individual components encountered in modern servers lead to a number of unknowns that are commonly in the range of several million. Therefore, efficient numerical methods and approximate models have to be developed.

**Related Publications: **

### Application of the Characteristic Mode Analysis to Antenna Design and Electromagnetic Compatibility

**Alexander-von-Humboldt Fellowship Prof. Qi Wu. 01.03.2014 – 29.02.2016**

The Characteristic Mode Analysis (CMA) – based on the fundamental works of Harrington and Mautz – is a well-known numerical tool for antenna analysis and design. Within the numerical framework of the method of moments (MoM) the characteristic modes are found by the solution of a weighted matrix eigenvalue equation formulated with the MoM system matrix. The resulting current distributions are the eigencurrents and eigenvalues, respectively. The eigenvalues indicate if the corresponding eigencurrents may store primarily magnetic energy, or electric energy, or if they are “at resonance” and may radiate efficiently if excited. The eigencurrents constitute a set of (orthogonal) modes characteristic to the investigated structure (hence the name) and independent of any excitation. In antenna design inspection of the eigencurrents allows to identify e.g. suitable locations for generators and ways to improve radiation efficiency. In EMC analysis, the goals are opposite: analyze a potentially radiating structure and reduce its potential for emission.

During the Alexander von Humboldt Fellowship of Prof. Qi Wu at TUHH CMA was applied successfully to antenna design and extended to EMC analysis of digital systems. A full report of his fellowship is given in the following document **PDF**.

Eigencurrents in A/m and radiation eigenpatterns of a cubic heatsink on a motherboard at 500 MHz obtained by CONCEPT-II. Image taken from: “Characteristic Mode Analysis of Radiating Structures in Digital Systems,” IEEE Electromagnetic Compatibility Magazine.

**Related Publications:**

### Accurate and Efficient Algorithms in the Method of Moments for the Analysis of High Intensity Radiated Field Coupling into Aircraft

**Ph. D. Thesis Arne Schröder. 01.09.2009 – 31.08.2013**

Bistatic radar cross section (RCS) of aircraft to the left. The comparison of two diffrent computation methods shows good agreement. (Source: TET, TUHH). (Source: TET, TUHH).

**Related Publications:**

### Development of a Full-Wave Module for the High Intensity Radiated Field (HIRF) Synthetic Environment

**EU-Project. 01.12.2008 – 13.05.2013**

The “HIRF Synthetic Environment”, financed by the EU with a total of 45 partners, research project had the goal to provide a numerical modelling framework to the aeronautic industry which can be used during the development phase of aircraft for prediction of adequate electromagnetic immunity. For this purpose several state-of-the-art computational electromagnetic codes were integrated into a common modeling platform and validated against each other and against measurements. From the part of the Institute of Electromagnetic Theory the MoM code CONCEPT-II could be successfully adapated and integrated into the synthetic environment.

In the future, the availability of such a HIRF Synthetic Environment framework becomes more and more significant in order to be able to cope with competing requirements of new designs, with the related increase of testing, and the general demand of costs and time reductions.

Surface current distribution on aircraft due to HIRF illumination calculated by CONCEPT-II (Source: TET, TUHH).

**Further Information:** http://cordis.europa.eu/project/rcn/89387_en.html