Abstract:Owing to the angle deviation of UAV patrol inspection image acquisition, the current detection method of transmission line tower tilt lacks positioning constraints, which leads to the need for repeated judgment of the transmission line tilt. To solve this problem, a method for detecting the inclination of transmission line towers based on the Beidou positioning signal and UAV inspection images is proposed. Constructing a transmission line tower tilt detection framework, the acquisition unit uses the Beidou measurement antenna installed on the tower to obtain its position information and the corresponding tower-monitoring image. After pre-processing by the monitoring unit, the collected information is transmitted to the data center through the communication network. The positioning module uses RTK positioning technology to solve the tower position in real time. The tilt detection module uses the YOLOv4 network to identify the tower target in the monitoring image, and regards it as the area of interest. The LSD algorithm is used to extract the central axis and calculate the tilt angle of the tower profile. Combined with the processing results of the positioning module, the tower tilt angle is monitored. The detection results are presented through the monitoring terminal, and safety warnings are provided. The experimental results show that this method can achieve precise positioning of tower positions, and the positioning results are very close to the actual situation. After 100 training sessions, the network loss can be reduced to 0.06, enabling tower tilt detection and determination of its inclination.

A Transmission Line Tower Tilt Detection Method Based on BeiDou Positioning Signals and UAV Inspection Images