Automatic Fault Detection of Power Lines using Unmanned Aerial Vehicle (UAV)

Abstract

Safety and automation are the two major challenges in the application of Unmanned Aerial Vehicle (UAV), commonly known as drone, to the power lines inspection and fault detection. While current state-of-the-art UAVs are equipped with collision avoidance features, there is less attention to the automatic and real-time fault detection of power lines using UAVs. This paper presents the architecture of three drone-oriented concept designs for automatic and real-time fault detection of power lines using UAVs. The proposed systems could be potential candidates for replacing traditional inspection methods of power lines, which are risky and costly. By incorporating a robust neural network, i.e., Artificial Intelligence (AI), and using appropriate and efficient sensors, the systems can automatically detect various faults and defects on power lines with high precision. We propose three concept design options comprised of different hardware/software components and their feasibility factors. For instance, FLIR Duo Pro R as a thermal sensor and Zenmuse XT for thermal vision have been proposed to be used in the concept designs. For data communication, the proposed designs use cloud-based virtual private network (VPN) for a secure connection between remote control (RC) of the UAV and the server. Based on the advantages and disadvantages of the three proposed design options, the most efficient design is also discussed. This design proposes a system with lightweight sensors, which could increase the flight time of the UAV. Further, the AI interface is coded on to the RC, making it economical, without any database for big data storage. The back-end of the neural network is stored in a cloud server. With the help of GSM antenna, the AI can run on the tablet if there is an available cellular network.