Research and application of flying-wing aircraft bow wave modeling method based on convolutional neural network

With the development of aviation technology, aircraft voyage improvement has become an urgent problem to be addressed. Aerial refueling technology has been widely used as an important method for improving endurance. In probe-and-drogue refueling (PDR), the tanker's hose-drogue system is offset by the bow wave generated by the receiver's head. By probe-and-drogue refueling, we mean soft aerial refueling, which is a method that transfers fuel by connecting the drogue of tanker with the probe near the oil receiver's head to transfer fuel. The ability to accurately calculate flow field changes caused by the bow wave and the hose-drogue system offset position have become key factors for successful docking during PDR. However, existing bow wave models based on the potential flow method cannot accurately calculate the bow wave flow field of a flying-wing aircraft. Therefore, this paper proposes a variety of bow wave models based on a convolutional neural network (CNN), which aims to accurately solve the bow wave flow field of a flying-wing unmanned aerial vehicle in the process of autonomous aerial refueling. In addition, a method to calculate the hose-drogue system position based on the dynamics of a multi-body system is proposed, which calculates the final position of the hose-drogue system after docking from the coupled flow field of the tanker and receiver. Finally, by applying the bow wave model to the position calculation of the hose-drogue system, it was verified that the CNN model can be accurately applied to research related to aerial refueling.