Aiming at the requirements of vision-based aerial refueling missions of unmanned aerial vehicles (UAV), a positioning and docking guidance simulation experiment of UAV aerial refueling cone is designed. An intelligent cone detection algorithm based on YOLOv4-tiny is proposed. The recognition and positioning of cone can be achieved in the case of changing scenario. A vision-based docking guidance law is designed to realize the docking guidance by adjusting the flight speed and the flight angle of the UAV. Experiment results show that the accuracy and speed of YOLOv4-tiny detection of the refueling cone can meet the requirements of positioning accuracy and speed in the autonomous docking process of the UAV. In both stationary and motion scenario, the use of vision-based guidance law allows the UAV to quickly track the refueling cone to complete the autonomous refueling process.