Achieving sustainable fusion energy critically depends on accurately modeling complex plasma dynamics within tokamak reactors, particularly in the Scrape-off Layer (SOL), where heat and particles directly interact with reactor walls, influencing reactor performance and component longevity. Particle-in-Cell (PIC) simulations, although highly accurate, are computationally expensive and time-consuming, limiting their use for iterative design and real-time control. We employ an Extreme Gradient Boosting (XGBoost)-based surrogate model to efficiently predict plasma potential along the tokamak SOL using data from PIC simulations under varying operating conditions. The machine learning (ML) approach integrates physics-informed segmentation of the spatial modeling domain, distinguishing sharply between sheath regions and the quasineutral bulk plasma. This segmentation substantially enhances the surrogate model’s predictive accuracy to localized physical phenomena, a marked improvement over traditional global modeling strategies. We utilize XGBoost regression with hyperparameter optimization achieved through a tailored leave-one-curve-out (LOCO) validation method, ensuring robust generalization to previously unseen plasma conditions. We found that a global model with segmented consideration of the spatial dimension based on boundary plasma physics captures localized behaviors more accurately. This segmented approach leads to a mean absolute percentage error (MAPE) of 3.2%, outperforming other methods. The main engineering application of this approach is the significant reduction in computational resources and simulation time required for fusion reactor design and real-time plasma control. This allows rapid iterative design, improved operational decision-making, and potentially extends the operational lifetime of reactor components. • Extreme Gradient Boosting-based machine learning accelerates plasma modeling in fusion reactors. • Extreme Gradient Boosting regression predicts the Scrape-off Layer plasma potential with 3.2% mean absolute percentage error. • The new surrogate model replaces computationally expensive Particle-in-Cell simulations. • Segmented spatial modeling improves accuracy in boundary layer physics. • The new approach enables real-time prediction for fusion design and control.

In the last two decades, data regarding engineering design and product development has increased rapidly. Big data exploration and mining offer numerous opportunities for engineering design; however, owing to the multitude of data sources and formats coupled with the high complexity of the design process, these techniques are yet to be utilised to the best of their full potential. In this study, a comprehensive assessment of the state-of-the-art data-driven engineering design (DDED) in the last 20 years was conducted. A scientometric approach was employed wherein first, a systematic article acquisition procedure was performed, where a dataset of 3339 articles related to engineering design and big data analytics applications were extracted from Web of Science (WoS) and Scopus. Thereafter, this dataset was reduced to a dataset of 366 articles based on concise data screening. The resulting articles were used to analyse the dynamics of research in DDED throughout the last 20 years and to detect the primary research topics related to DDED, the most influential authors, and the papers with the highest impact in the DDED domain. Furthermore, the co-occurrence network of keywords/keyphrases and co-authorship networks were constructed and analysed to reveal the interconnection of the research topics and the collaboration between the most prolific authors. Finally, an insight how big data analytics is being applied through product development activities to support decision-making in engineering design was presented.