Artificial Intelligence (AI) has become essential to various industries, from entertainment to autonomous driving, providing innovations that streamline processes and improve efficiency. With advances in machine learning (ML) and data-driven technologies, AI is now taking center stage in more specialized fields. Still, it was not until recent years that AI started to rise in EMC, and it has yet to have a drastic impact as in other areas. Historically, EMC engineers faced significant challenges in ensuring that electronic devices functioned without interference, and in recent times, these challenges have increased exponentially [1]. Then it would only make sense that AI is leveraged to aid in many of these challenges.
One of the first applications of AI specifically aimed at solving EMC problems appeared in 2010 [2] where AI was used to predict emissions from PCBs and cables within EMC chambers. Since 2020, the use of AI in EMC has increased, with many innovative approaches being developed to improve design, testing, and troubleshooting in the field. Here are a few of the AI applications in EMC.
AI as an Assistant in the PCB Design Process
One of the most common applications of AI in EMC is its role in assisting with the design of electronic devices, particularly printed circuit boards (PCBs). Designing a PCB that adheres to EMC best practices can become complex with large systems, as it requires minimizing the possibility of electromagnetic interference (EMI) while ensuring minimum functionality. AI-driven tools can aid engineers by recommending optimizing all the stages during the design of PCB layouts, suggesting placement of components, and even selecting appropriate materials to aid in achieving EMC.
Many AI applications in EMC have fallen so far on the signal and power integrity (SIPI) field. For example, [3] presents the idea of creating a workbench to aid the implementation of AI in SIPI-related applications. One of the presented examples focuses on aiding the design and implementation of decoupling capacitors in PCBs. The AI algorithms in this instance predicted the optimal locations for these capacitors to maximize EMC performance. Although AI algorithms and architectures are being developed, one of the main limitations is the lack of databases to train and test models. Fortunately, this has been noticed and some databases have started to appear to solve this issue [4].
AI and Computational Electromagnetics
AI is also making significant contributions in the field of Computational Electromagnetics (CEM), where traditional methods for simulating electromagnetic fields can be computationally expensive and time-consuming. AI, particularly ML techniques, has been applied to speed up these simulations while trying to not sacrifice accuracy.
An example is the work described in [5]. In this research, machine learning models were trained on data from CEM simulations to create predictive algorithms that could model complex electromagnetic interactions. These models can allow to run simulations more efficiently, enabling faster iterations in the design process. This is particularly useful in SIPI analysis, where CEM traditionally requires significant computational resources. The AI-driven methods proposed in the paper not only improve simulation times but also ensure more robust results, aiding engineers in developing designs quicker than before. Not fully sorted out for EMC.
AI in Broader EMC Applications
Beyond PCB design and CEM simulations, AI has shown great potential in other critical areas of EMC, particularly in optimizing testing and diagnosing EMC issues. Here in ETERNITY AI is being utilized for predictive EMC. A recent paper [6] demonstrated the use of neural networks to reconstruct electromagnetic fields. By analyzing samples and the location of important factors in scenarios, the AI model provided predictions of the spatial distribution and the intensity of electric fields in simple scenarios. This study shows the potential of AI in reconstructing electromagnetic field behavior.
The Future of AI in EMC
AI’s role in EMC is expanding rapidly, and the future looks promising. New developments, such as AI-driven tools for automated testing, and real-time interference mitigation are expected to come. One initiative to push the use of AI in EMC is the European project PATTERN. The project focuses on developing AI-based tools to assess electromagnetic environments, define AI tools for aiding design, and provide real-time feedback on EMC issues, ultimately speeding up the development process and aiding the process of applying risk-based EMC. For more information, you can visit the project page.
In conclusion, AI has started to permeate EMC by providing powerful design, simulation, and testing tools. As AI continues to evolve, its applications in EMC will only grow, helping engineers address the increasing complexity of modern electronic systems while meeting the EMC requirements. With projects like PATTERN leading the way, the future of AI in EMC looks brighter than ever.
References
[1] S. R. Veléz, M. J. A. M. van Helvoort, R. Vogt-Ardatjew, and F. Leferink, “The Need for EMI Risk Management in MRI Systems,” in 2023 IEEE 7th Global Electromagnetic Compatibility Conference (GEMCCON), 2023, pp. 60–61. doi: 10.1109/GEMCCON57842.2023.10078222.
[2] G. McCormick, Z. A. Khan, V. Devabhaktuni, M. Alam, and A. Wood, “Estimating radiated emissions from printed circuit boards and cables inside EMC chambers,” in 2010 IEEE International Symposium on Electromagnetic Compatibility, 2010, pp. 36–39. doi: 10.1109/ISEMC.2010.5711243.
[3] W. John, E. Ecik, N. G. Shoaee, J. Withöft, R. Brüning, and J. Götze, “AI Workbench – Conceptual Workflow to Develop AI Models for SI/PI-Applications in PCB Development,” in 2024 IEEE Joint International Symposium on Electromagnetic Compatibility, Signal & Power Integrity: EMC Japan / Asia-Pacific International Symposium on Electromagnetic Compatibility (EMC Japan/APEMC Okinawa), 2024, pp. 233–236. doi: 10.23919/EMCJapan/APEMCOkinaw58965.2024.10585118.
[4] M. Schierholz et al., “SI/PI-Database of PCB-Based Interconnects for Machine Learning Applications,” IEEE Access, vol. 9, pp. 34423–34432, 2021, doi: 10.1109/ACCESS.2021.3061788.
[5] L. Jiang, H. Yao, H. Zhang, and W. E. I. Sha, “Developing Machine Learning CEM Methods for EMC/SI/PI”.
[6] S. M. S. Laurens, J. D. Bertaux, and A. Roc’h, “Neural Network for the Prediction of Electric Field Intensity Applied to a Simple Scenario,” in 2024 IEEE Joint International Symposium on Electromagnetic Compatibility, Signal & Power Integrity: EMC Japan / Asia-Pacific International Symposium on Electromagnetic Compatibility (EMC Japan/APEMC Okinawa), 2024, pp. 702–705. doi: 10.23919/EMCJapan/APEMCOkinaw58965.2024.10584958.