Slovenia
University of Ljubljana researchers from the Faculty of Mechanical Engineering have developed a machine learning based approach that significantly accelerates complex plasma simulations used in fusion energy research. The innovation helps predict plasma behavior much faster, supporting the development of more efficient and stable fusion reactors as a future source of clean energy.The research focuses on plasma dynamics in the scrape-off layer of fusion devices, which is traditionally simulated using highly computationally intensive particle-in-cell methods. While accurate, these conventional simulations require large amounts of computing power and time, limiting the ability to explore different operational conditions.To address this challenge, researchers including Nikola Vukašinović, Leon Kos, Uroš Urbas, and Ivona Vasileska developed a machine learning based surrogate modeling approach using XGBoost algorithms. These models learn relationships between input parameters and simulation outputs, enabling fast and accurate predictions of plasma properties such as electric potential at the divertor target.The results show that the new method significantly reduces simulation time while maintaining high accuracy, making it a powerful tool for advancing fusion research.According to the researchers, this approach opens new possibilities for analyzing complex plasma processes and could contribute to faster development of fusion reactors in the future. The study has been published in Engineering Applications of Artificial Intelligence.
Source: University of Ljubljana
#WordMain #StudentNewsPortal #Europe #studentnews
About the Author
