Expansion of the EPED-NN database for DIII-D RMP ELM suppressed discharges

In this project, we expand the RMP EPED database with information on plasma rotation at the top of the pedestal and SURFMN predictions for islands where external field penetration is predicted by resistive plasma response models. Resonant Magnetic Perturbations (RMPs) control dangerous Edge Localized Modes (ELMs) in H-mode discharges by limiting the H-mode pressure pedestal growth below the critical level. Machine learning models, particularly neural networks (NN), offer a data-driven approach for pedestal predictions. EPED is the leading model to describe and predict the pedestal height and width, and an accelerated NN version was built on a large database of EPED simulations. Previous research has shown that EPED and EPED-NN over-predict the height and width of the pedestal in RMP ELM-suppressed discharges. This project builds on the OMFIT framework to improve a vast database of pedestal parameters, including their uncertainties in DIII-D RMP ELM-suppressed discharges. Automated pedestal predictions for new plasma discharges will guide RMP ELM suppression scenario development towards high pedestal regimes while suppressing ELMs.