Review of model-driven and data-driven/AI approaches for estimating frequency dynamics after a disturbance

The increasing integration of renewable energy sources (RESs) through power electronic converters and inverters (PECIs) in modern power systems has led to a consistent decrease in system inertia, which poses significant challenges for managing and stabilizing complex frequency dynamics. Consequently, maintaining frequency stability and ensuring the security of the power grid have become critical issues. Addressing these issue requires the development of appropriate methods that accurately capture the complex frequency dynamics introduced by RESs. This review comprehensively surveys the changes in frequency dynamics following the integration of RESs and focuses on modeling, analysis, and application. The basic properties of system inertia, variations in inertia contributions across different power systems, effects of RES integration in different frequency-response stages, and current mainstream methods for estimating frequency dynamics are reviewed systematically. Essential data, algorithms, and software requirements are analyzed and a hybrid framework based on Improved System Frequency Response–Multi-scale Convolution Layer Improved Transformer (ISFR-MCIT) that integrates model-driven and data-driven approaches to enhance estimation accuracy in low-inertia systems is proposed. Case studies reveal that the proposed methods achieve a mean absolute error (MAE) of 0.0006 Hz for frequency nadir and 0.0016 s for nadir time validated in the Chinese Society for Electrical Engineering - Frequency Stability (CSEE-FS) test system, outperforming existing models by up to 60 % in accuracy metrics under varying scenarios. Finally, an outlook is provided on the quantitative assessment of frequency security and response, alongside the development of an equivalent inertia assessment system for future research. • Review of frequency dynamics estimation methods in renewable energy sources (RES) dominated power systems. • Hybrid algorithmic approaches proposed for frequency stability in low-inertia power grids. • Strategic recommendations for quantitatively assessment of frequency security and response.