Francesca Rossi, Mauro Garcia Lorenzo, Eduardo Iraola de Acevedo, Elia Mateu Barriendos, Vinicius Albernaz Lacerda, Francesc Lordan–Gomis, Rosa Badia, Eduardo Prieto–Araujo
• An open-source HPC-enabled framework generates stability datasets for IBR-rich grids. • A scalable operating-space formulation reduces dimensionality while retaining key dynamic features. • The operating space captures the effects of IBR penetration, control tuning, and GFM/GFL control roles. • Adaptive, sensitivity-guided and entropy-based sampling targets the small-signal stability boundary. • Machine-learning models trained on the data reach up to 92% stability accuracy.