A Review of Artificial Intelligence Techniques for Ferroresonance Detection and Mitigation in Power Systems

Ferroresonance is a nonlinear phenomenon in power systems that can produce irregular oscillations and severe overvoltages, leading to insulation stress and damage to transformers, voltage transformers, cables, and associated equipment. The increasing penetration of renewable energy sources, inverter-based distributed generation, underground cables, and complex grounding configurations has expanded the operating conditions under which ferroresonance may occur in modern grids. Conventional analytical methods and protection schemes often exhibit limitations in representing nonlinear magnetization, hysteresis effects, chaotic behavior, and multimodal resonance responses. This paper provides a comprehensive review of ferroresonance detection and mitigation techniques reported till 2025 with a focus on last five years. Numerical modeling approaches, electromagnetic transient simulation tools, ferroresonance modes, and conventional mitigation strategies are systematically examined. Particular emphasis is placed on recent applications of artificial intelligence, including machine learning, deep learning, fuzzy logic systems, evolutionary algorithms, expert systems, and hybrid intelligent frameworks. A comparative analysis is presented to evaluate these methods in terms of detection accuracy, computational complexity, interpretability, and suitability for real-time protection. The review highlights the complementary roles of data-driven intelligence and physics-based modeling and emphasizes integrated approaches as a practical pathway toward improving reliability, protection performance, and resilience in evolving smart grid architectures.