The global integration of renewable energy is fundamentally reshaping power systems, replacing synchronous generators with inverter-based resources (IBRs). This transition introduces a critical stability challenge: conventional Grid-Following (GFL) inverters, which depend on a stable grid voltage for synchronization, become unstable in weak grids with low Short-Circuit Ratios (SCR). This paper provides a comprehensive review of the paradigm shift from GFL to Grid-Forming (GFM) inverters—the key to stabilizing future power systems. We analyze the core control architectures, operational limitations, and stability mechanisms of both paradigms, with a focused examination of synchronization instability in GFL units, including a precise small-signal impedance model explaining the negative incremental resistance effect and its impact on critical SCR thresholds in multi-inverter systems. We further delve into the voltage-source behavior of GFM strategies like Droop Control and the Virtual Synchronous Machine (VSM), critically comparing their power sharing and frequency regulation capabilities in islanded microgrids and analyzing the stability trade-offs of virtual inertia in ultra-weak grids. The review covers advanced mitigation techniques, including adaptive control and virtual impedance, while also addressing their limitations in P-Q decoupling in resistive grids and exploring novel control paradigms. We provide a bifurcation analysis of large-signal instabilities, clarifying the role of current limiting strategies in catastrophic failures. Furthermore, we highlight the transformative role of Artificial Intelligence (AI) in enabling real-time, self-tuning inverters, and critically examine the optimal role of the AI layer (supervisory vs. direct control) and the attendant risks. Finally, we present a forward-looking research trajectory for developing an AI-augmented GFM inverter capable of stable operation in ultra-weak grids (SCR < 2), including the crucial development of Verification and Validation (V&V) frameworks for AI-based grid controllers to provide formal or probabilistic guarantees against adversarial data and cyber-physical attacks, providing a clear pathway to overcome the most pressing stability barriers in renewable-rich power networks. • Provides a quantitative stability-oriented framework for the transition from grid-following to grid-forming inverters. • Analyzes synchronization and large-signal instability mechanisms in low-SCR and ultra-weak grids. • Critically compares droop and virtual synchronous machine grid-forming strategies under weak-grid conditions. • Examines adaptive, impedance-based, and AI-augmented control approaches with stability-theoretic insights. • Discusses verification, validation, and robustness challenges for AI-enabled grid-forming controllers in future power systems.
Abdelwahab et al. (2026) studied this question.