AI-Driven Predictive Maintenance for Sustainable Power Grid Infrastructure

The operation of modern industrial facilities relies inevitably on a stable electricity supply, necessitating continuous monitoring and timely preventive or corrective maintenance interventions. Rapid advancements in information and communication technologies (ICT) and artificial intelligence (AI) technologies are transforming energy production, transmission, and distribution processes, aligning with sustainable development goals and the green transition.

This study analyses the current state of AI applications in monitoring the integrity of power grid infrastructure. A significant potential is identified in AI tools, such as convolutional neural networks (CNNs), for detecting potential faults or failure locations from sequences of photographs or video footage of electrical facilities. By comparing existing AI solutions like YOLO for object detection and anomaly detection models, the study evaluates required input data formats (e.g., high-resolution imagery) and valuable algorithmic outputs, including fault probability scores and localization heat maps. Through discussion and analysis of these approaches, the study clarifies future research directions, identifies remaining gaps, and proposes sequential algorithms, from data pre-processing to real-time deployment, for enhanced maintenance efficiency.

Future research should focus on resource optimization, enabling faster decision-making in energy sectors and contributing to environmental protection as key pillars of industrial engineering in the era of artificial intelligence. This includes reducing waste through predictive interventions and facilitating integration of renewable energy sources (RES), supporting the green transition toward a competitive, low-carbon economy.