Research On Trajectory Prediction Method Of Small UAV Based On Trackdata

With the development of UAV technology,the operation of a large number of UAVs poses new challenges for air traffic management.According to the current unsafe incident of unmanned aerial vehicles,the core of the UAV air traffic control is the collision avoidance problem,and the trajectory prediction is the key technology to solve the collision avoidance problem.At present,the research on trajectory prediction at home and abroad mainly has problems such as low precision,difficult to obtain parameters,and accumulated error.Aiming at the shortcomings of the current research,this paper proposes a small UAV track prediction method based on the track mode,which is proposed for the small UAV ground decision system.Firstly,the trajectory of the UAV has certain changes in different modes.Therefore,this dissertation studies the identification method of the trajectory mode of the small UAV.According to the actual operation research of small unmanned aerial vehicles and comprehensive research at home and abroad,six kinds of trajectory patterns are defined.A set of trajectories is generated to establish kinematic indicators and statistics,so as to establish a small UAV trajectory feature space and perform feature selection to reduce computational complexity and improve accuracy.On the basis of summarizing the existing multi-classification methods,a multi-classification method based on improved binary tree is proposed,and the accuracy of the method is checked by simulation data and real flight data.Among them,various trajectory pre-processing methods are compared for real flight data of UAVs.With simulation data and real flight data verification,the classification accuracy of the track pattern recognition method reaches 80%Secondly,due to the problem that the parameters of the classical trajectory prediction method are difficult to obtain,based on the trajectory pattern recognition,a small UAV trajectory prediction method based on neural network is proposed to bypass the wind tunnel test and Parameter identification.The small UAV trajectory prediction model is established by BP neural network.The dynamics and kinematics model of small UAVs are analyzed,and the UAV system model is simplified and improved under certain conditions.The prediction accuracy of the method is analyzed by collecting simulation data of small UAVs at different wind speeds and expected speeds.From the limited results,the prediction method error is within 2m.Finally,the application of trajectory prediction technology in UAV collision detection is studied.Based on the current definition of UAV protection area,the experimental scenario for horizontal maneuver problem in the case of non-cooperative target intrusion is designed.Based on the trajectory prediction,the conflict detection index is proposed,and the mobile reference system is established and solved by simulation experiment.Analyze.The requirements and early warning requirements for conflict detection sensors are presented.