Effective stability analysis is essential for the secure operation of modern power systems. As smart grids evolve with increased interconnection, renewable energy integration, and electrification, the large-scale deployment of ultra-high voltage AC/DC networks introduces various operational modes and potential fault points, posing significant challenges to maintaining stability. Traditional analysis and control methods fall short under these conditions. In contrast, emerging artificial intelligence (AI) techniques, combined with real-time data collection, provide powerful tools for enhancing stability analysis in smart grids. This paper comprehensively explores AI techniques in stability analysis, discussing the necessity and rationale for integrating AI into stability analysis through the lenses of knowledge fusion, discovery, and adaptation. It provides a thorough review of current studies on AI applications in stability analysis, addresses key challenges, and outlines future prospects for AI integration, highlighting its potential to improve analytical capabilities in complex power systems.

This paper presents a novel Koopman Operator based framework to estimate the region of attraction for power system transient stability analysis. The Koopman eigenfunctions are used to numerically construct a Lyapunov function. Then the level set of the function is utilized to estimate the boundary of the region of attraction. The method provides a systematic method to construct the Lyapunov function with data sampled from the state space, which suits any power system models and is easy to use compared to traditional Lyapunov direct methods. In addition, the constructed Lyapunov function can capture the geometric properties of the region of attraction, thus providing useful information about the instability modes. The method has been verified by a simple illustrative example and three power system models, including a voltage source converter interfaced system to analyze the large signal synchronizing instability induced by the phase lock loop dynamics. The proposed method provides an alternative approach to understanding the geometric properties and estimating the boundary of the region of attraction of power systems in a data driven manner.

High-frequency resonance can occur when a modular multilevel converter (MMC) is inserted into an AC grid. Additional damping control is a relatively low-cost resonance suppression strategy compared to passive damping strategies. This paper analyzes the influences of a feed-forward voltage filter and feedback current filter for the inner controller for the high-frequency impedance characteristics of the MMC based on a model. Moreover, the mechanism, influencing factors, and limitations of the existing strategy including an additional low-pass filter in the voltage feed-forward stage are investigated. Secondly, a resonance suppression strategy for the inclusion of additional cascaded notch filters in the voltage feed-forward stage is proposed, and its parameter design method and applicable scenarios are analyzed. In addition, this paper analyzes the effects of the inclusion of an additional control in other stages for the inner controller of the MMC. Finally, the correctness of the theoretical analysis and the proposed strategy is verified based on the simulation of an actual project on PSCAD/EMTDC.