IEEE EMC-S PL reported activity no. 04/2025: "IEEE EMC Society Distinguished Lecturer - prof. Xing Chang Wei"

#EMC #emc

Wykład 1: “Can You Identify an Electromagnetic Photo? – EMC Analysis Enhanced by Artificial Intelligence” godzina 10.00 do 11.30

Wykład 2: “Practical EMI Measurements and EMI Probes Design” godzina 12.00 do 13.30

Wykłady z serii „IEEE EMC Society Distinguished Lecturer Program” finansowane są przez „centralę” IEEE EMC Society (Towarzystwo Kompatybilności Elektromagnetycznej) z USA i odbędą się w siedzibie Łukasiewicz - Poznańskim Instytucie Technologicznym, ul. Estkowskiego 6, Poznań

Talk 1: Can You Identify an Electromagnetic Photo? – EMC Analysis Enhanced by Artificial Intelligence
Abstract: In recent years, the artificial intelligence (AI) technology provides a powerful tool for solving electromagnetic problems, and there has been many successful stories for their applications on microwave device and antenna designs. The radiated near-field can be taken as an electromagnetic photo of an unknown EMI radiation source. This photo contains a lot of intrinsic information about the radiation source, such as its 3-meter radiation, coupling characteristics with nearby sensitive devices, as well as information about the position and polarization of the radiation source itself. But due to the inability of the human eye to see electromagnetic waves, our ability to identify electromagnetic photos is much lower than that of ordinary photos. AI has achieved significant results in facial recognition. This allows us to use AI to process electromagnetic photos and extract the useful information for EMI analysis from the features of the photos, such as 3-meter far-field and far-field radiation pattern.

This talk will start with a brief overview of AI and its applications in the EMC area. Then several different ways to enhance the near-field scanning by AI are presented. The Green’s function hybrid with artificial neural network (ANN) is developed for EMI estimation. The Green’s function of a dipole array with fixed source points is taken as input and the radiated field at any given observation point is taken as the output of ANN. We use the powerful mapping ability of ANN to replace the matrix-vector multiplication between Green’s function and dipole moments in the traditional dipole model, so that the ANN can be used to predict the near-field from unknown EMI source. Next, a deep convolutional neural network (DCNN) hybrid with the plane wave spectrum is proposed. By leveraging plane wave expansion, the spatial magnetic near-field data are converted into the spectrum domain, serving as the input for the DCNN model. DCNN’s output is the 3-meter electric far field. It enables the output of DCNN (3-meter far-field) insensitive to variations in the near-field scanning height. Finally, the physics-informed neural network (PINN) is introduced for near-field prediction, where the wave equation is integrated with the deep neural network. Therefore, the PINN is capable of efficiently interpolating and extrapolating the scanned near-field fields.

Talk 2: Practical EMI Measurements and EMI Probes Designs
Abstract: Electromagnetic interference (EMI) modeling and debugging require different suitable measurement methods. For high-speed circuits and integrated circuits, the IEC published a series of radiated emission and immunity testing methods.

In this talk, we will briefly introduce EMI measurement methods, including the TEM/GTEM, surface scan, 1/150 Ohm method, reverberation chamber, and anechoic chamber. Next, we will focus on the practice of designing electric and magnetic probes for EMI testing. What EMI testing requires is not a simple loop or monopole antenna as a suitable probe. In practical applications, one should consider probe’s sensitivity, effective center, spatial resolution, bandwidth and immunity to unwanted field components. A compact multi-components EzHxHy probe is proposed. In comparing with available single component probes, the EzHxHy probe can greatly save the engineer’s time. Furthermore, a wideband EzEx probe is proposed through the integration of a strip line magic-T. The magic-T can effectively streamline the complex data processing, enabling the proposed probe to offer a flexible instrument selection and achieve high measurement efficiency. After that, an Ex probe covering a wide frequency band (1.4–7.0 GHz) based on common mode absorbing is presented, where the self-resonance of a traditional Ex probe is eliminated by an absorbing-type balun. Meanwhile, other probes including the high spatial resolution Ez probe and high frequency (up to 70GHz) Hx probe will also be discussed.

• Add Event to Calendar
  iCal
  Google Calendar

• Lukasiewicz-PIT
• Estkowskiego 6
• Poznan, Wielkopolskie
• Poland 61-755

---

• Contact Event Host
• Co-sponsored by Lukasiewicz - PIT, POLLAB

• Link to External Registration