PREDICTIVE MAINTENANCE IN POWER TRANSFORMERS: A SYSTEMATIC REVIEW OF AI AND IOT APPLICATIONS

Abstract

Power transformers are critical assets in electrical power systems, and their failure can result in costly downtime and catastrophic grid disruptions. This systematic review investigates the emerging role of Artificial Intelligence (AI) and Internet of Things (IoT) technologies in enabling predictive maintenance (PdM) of power transformers. Drawing upon 126 peer-reviewed articles published between 2015 and 2024, this review categorizes and synthesizes state-of-the-art techniques involving sensor integration, real-time condition monitoring, data fusion, machine learning (ML), deep learning, and digital twin frameworks. The analysis reveals a growing trend toward hybrid PdM models that leverage transformer health indices, vibration and thermal imaging, dissolved gas analysis (DGA), and partial discharge (PD) data. Neural networks, support vector machines, decision trees, and ensemble methods dominate the AI approaches, while IoT-based sensor networks and cloud-edge computing architectures underpin the system infrastructures. Key challenges identified include data heterogeneity, cybersecurity vulnerabilities, high initial costs, and lack of standardization in deployment practices. This review concludes that integrating AI and IoT in transformer maintenance not only enhances fault detection and failure prediction but also supports asset lifecycle optimization and grid resilience. The findings contribute to academic research and industrial applications by providing a consolidated framework for future development, standardization, and policy formulation in smart grid systems.