Research on UAV Autopilot and Tower Identification Technology Based on Cloud Control Collaboration

With the increasing popularity of drone inspection technology in the power grid industry, the old lines of the power system continue to increase, the line hazards are increasing day by day, and the bird damage is extremely serious. In order to solve the problem of high professional requirements of inspection personnel for UAV inspection, it is impossible to quickly find potential risks of the line in the inspection process in time. The distribution network identification algorithm model, insulator aging identification algorithm model, lightning arrester aging identification algorithm model and bird damage identification algorithm model are lightly transformed to achieve the quantification, pruning and distillation of the above models. Through the lightweight front-end recognition algorithm, the UAV can realize auto-pilot, and in the inspection process, it can assist inspectors to quickly locate and find the problems of insulator aging, lightning arrester aging and bird damage, and then complete the elimination of defects in a timely manner, so as to avoid operational safety problems caused by defects of the old lines and to ensure the stable operation of the power grid.