Trajectory Prediction Of High Maneuvering Target

Tracking and prediction of the trajectory of a target with high maneuvering ability is one very important ?eld of estimation, and it is widely applied in a lot of important ?elds such as military surveilance, civil planes control and soon. The precision of tracking and prediction is usually deteriorated because of the unstopping change of motion models of the target and the nonlinearity of the motion model and the measurement model. In this master thesis, study for the problems about maneuvering target tracking has been conducted, and one high maneuvering target tracking algorithm based on multiple model methods is studied. In this paper, the motion models have been studied ?rst, from the un-maneuvering motion model, axis uncoupled maneuvering target model, to the three dimensional turn model. And then the choice of coordination system and corresponding ?lter has been studied. In the procedure of designing and application of multiple model methods, the members of motion model set, coordination system and ?lter should be chosen carefully. In order to testify and analyze the performance of multiple model methods, one trajectory for simulation is designed at the very ?rst place, and it is used as the benchmark for further comparison of different algorithms. In this experiment, one assumption that no matter how large the error of tracking or prediction, the target remains in the sensor scope is made. Three motion models including constant velocity model, constant acceleration model, and three dimensional turn model has been included in the motion model set in order to avoid the amount of calculation of being too large. The mixed measurement model is chosen to remain the white character of the measurement noise, which is the target state remains in the Cartesian coordination system and the measurement state vector is in sensor coordination system. To select the basic ?lter for constant velocity, and constant acceleration motion model, standard Kalman ?lter is decided. And for the turn model, extended Kalman ?lter is chosen. By comparing the difference of the true trajectory and the prediction one, it is obvious that the multiple model methods increase. After studying the working conditions of the large aperture telescope which can be transported, we got the design requirements of the shafting structures.We carried on the analysis to the parameters which can affect the bearing capacity of plane thrust bearings, angular-contact ball bearings and cylindrical roller bearings by using the finite element analysis models and got the design parameters of the shafting structures which meet the requirements of the subject. The results showed that the plane thrust bearing can meet the design requirements, and it was suggested to use this type of bearing. In the future, if the shafting structure we designed needs to carry larger aperture telescope or stronger impact load and the plane thrust bearing can’t meet the design requirements, we propose the use of angular-contact ball bearings or cylindrical roller bearings. precision of the tracking and prediction ability. At last, image-enhanced method to further improve the performance of the multiple model methods have been studied. Normally, the point information about the target is given but information about attitude and orientation has just been ignored. In this paper the way importing such information into the multiple model method is studied.