Trajectory Estimation Of Hypersonic Target Based On Motion Behavior Recognition

In the modern military warfare,hypersonic weapons are very promising tool for fighting The United States and other countries have been concentrating on their efforts to develop it.The research of China on hypersonic targets is still in the initial stage.Recently,how to defense the hypersonic weapons has become an important military issue.The aim of a trajectory estimation problem is to get the target location,speed and acceleration information.It is the key of defense.In the traditional hypersonic target trajectory estimation process,as the target movement cannot be obtained,the aerodynamic estimation is usually taken into account in the filter,which leads to the error of estimation results usually very large.In this thesis,a trajectory estimation method based on motion behavior recognition is proposed to improve the effect of trajectory estimation by combining the machine learning method with the Extended Kalman Filter technique.The main research work is as follows.Firstly,the kinematic model of hypersonic targets is established.The coordinate system and its conversion relation in the process of motion are described,the kinematic model is deduced,and the kinematic model is simplified.Secondly,the movement characteristics of the target are analyzed and the movement behavior is described as parameterized model.Based on the analysis of longitudinal and lateral motion characteristics,typical longitudinal motion behaviors are obtained,including quasi-equilibrium glide and constant-attack-angle skip;lateral motion behaviors include no maneuver,weave maneuver and turn maneuver.A parametric model is established for the Quasi-balanced gliding process.Meanwhile,a parametric model of lateral aerodynamic forces is established for swing maneuver and turning maneuver as well.Thirdly,a method of moving behavior recognition based on machine learning is proposed.The train data processing method is given for the target behavior identification classifier.The longitudinal and lateral motion behavior identification classifiers are designed based on Random Forest and Adaboost to identify the longitudinal and lateral motion behaviors respectively.Through the analysis and comparison of the classification results,the recognition method of different motion behaviors is obtained.Finally,a hypersonic target trajectory estimation method is proposed based on motion behavior recognition result obtained via classifiers.The aerodynamic model is obtained by identifying the moving behavior of the target by the classifier.Then an EKF filter is used to estimate the position and velocity information of the target.In the process,the dynamic estimation of the target trajectory is realized.The simulation results show the feasibility and effectiveness of the proposed trajectory estimation method.