Electricity infrastructure inspection using AI and edge platform-based UAVs

Abstract

Traditional electricity infrastructure inspections usually have high costs, risks and it takes a long time for specialized personnel to carry them out. Additionally, they also involve scaffolding risks, that lead to a high accident rate in most electricity companies. The recent emergence of Unmanned Aerial Vehicles (UAVs) is gradually leveraged to avoid such risks. However, UAVs usually face Global Positioning System instability issues especially in the distant or harsh infrastructure areas. This requires frequent manual UAV control and calibration by electricity operators. In this article, we propose a new method for automating the UAV infrastructure inspection procedure. The method uses Artificial Intelligence techniques to identify electricity infrastructures and the associated assets, as well for the real-time detection of infrastructure faults. Additionally, using 5G Network Function Virtualization technologies, such as end-to-end network slicing, combined with edge computing, significant latency and GPS accuracy improvements are realized during the inspection. We apply the method for the inspection of a Hydroelectric Power Plant of the Public Power Corporation. The experiments illustrate significant benefits in latency, GPS accuracy, fault discovery rate and accident reduction. Such benefits provide real-time response to infrastructure faults that supports business continuity and increase customer satisfaction.

Keywords

• Unmanned Aerial Vehicles
• 5G
• Multi-access edge computing
• Network slicing
• Electrical fault detection
• Network security monitoring