Power line monitoring

Insulator heights and wire slack. Checking the accuracy.

1. Power line inspection with drones

The survey and inspection of power lines with unmanned aerial vehicles have become quite widespread. Survey data are used for cadastre, engineering and surveying. Quite often the question arises: is it possible to obtain elevation marks of power lines, sags, wire dimensions and other metric characteristics using a UAV? And with what precision?

The main sensor of the UAV is a camera. The 3D image is built from overlapping photographs and information about the exact centers of photography. But as always, there are nuances. After processing a set of images of power transmission lines in photogrammetric software, you will not see supports and wires in 3D. The most that you can get is a dense cloud. Unfortunately, only a few of the many supports will be reconstructed properly. See example below:

Power lines are too thin to be reconstructed in 3D by photogrammetric software. The algorithm simply can't detect the long "thread" several pixels wide. In this case, computer vision comes to the rescue and there are multiple technologies for automated wire detection. Despite not having a large team of programmers, we have developed a technology for accurately detecting and reconstructing the position of power lines using standard tools. This allows us to obtain cable and ground height in any location with high precision, and to tackle related tasks such as determining wire dimensions, sag, turning angles etc.

2. Control measurement of wire heights

In order to check the accuracy of the wire height detection we surveyed a small section of a power line. First, we set up ground control points and obtained their coordinates using the EFT M3 GNSS in RTK mode. Using the AR600E high-voltage cable height meter, we measured the height of the bottom wire at each control point.

This way we collected the control wire measurements using the instrumental method. Next, we conducted aerial survey of the site and determined the height of the wire above the control point using the photogrammetric method. Last step was to compare the instrumental measurements with the aerial survey based data. The aerial survey was performed with a Phantom 4 Pro v2 PPK survey grade quadcopter from uav-design.com

3. Data processing and accuracy assessment

Coordinates of image centers were processed in Magnet Tools software. Photogrammetric processing was carried out in Metashape. Images were captured from 100 meters altitude in two passes. Survey technology with a geodetic drone from uav-design allows to consistently obtain high accuracy results. For our flight, we got average camera alignment error within 2 cm. The smaller the error in the relative position of the cameras, the more accurate the wire position will be determined:

The root mean square error at ground control points was also several centimeters:

Using our own technology, the power line heights were determined at the control point locations. A total of 5 measurements were taken with the photogrammetric method at different sections of the wire.

The final table below shows the difference in the height of the wires measured by the instrumental and photogrammetric method. The maximum deviation was 12 cm, the average deviation was 6 cm.

Conclusion: Photogrammetric survey of power lines with drones using the technology developed by Aerostream makes it possible to obtain wire positions with an accuracy of about 6-12 cm.