Research On Adaptive Multiple Maneuvering Extended Target Tracking And Trajectory Estimation Method

Target tracking technology is closely related to people’s production and life,military and aerospace systems,and has rich academic research significance and high application value.The problem of multiple extended target tracking arises with the development of modern highprecision sensing devices,attracting the research interest of scholars at home and abroad.With the rapid development of this technology,it also faces many challenges,such as: target maneuvering,target number change,target trajectory estimation and so on.In this paper,under the background of random finite set theory,based on the extended target probability hypothesis density(ET-PHD)filtering framework,the research on multiple maneuvering extended target tracking and trajectory estimation methods is carried out.The main work is as follows:(1)Considering the problem that the traditional multiple extended target PHD filter is difficult to adapt to maneuvering target tracking,an adaptive parameter particle PHD filter for multiple maneuvering extended target tracking is proposed.The algorithm uses an adaptive target maneuvering parameter estimation method,which can jointly estimate the target state and target maneuvering parameters;and improves the particle filter resampling process of the algorithm,which can adaptively detect degraded particles and re-collect particles in highlikelihood regions of the target state space to improve sample diversity,enabling the algorithm to provide higher accuracy when fitting the target state.In addition,for the identification of new targets,a measurement-driven technology is proposed to automatically identify and track new targets.The experimental results show that the algorithm can effectively track multiple maneuvering extended targets in complex scenes,has strong adaptability to system noise and clutter,and has good stability.(2)Considering the tracking problem when the target dynamic model is unknown in complex scenes and lack of prior information such as the distribution of new targets,an adaptive grid-driven PHD filter for multiple maneuvering extended target tracking is proposed.The algorithm can recursively estimate the target state under the premise that the distribution of the new target and the dynamic model of the target are unknown.The algorithm uses the dynamic trend of the grid to respond to the unknown dynamic model of each target.It can adaptively adjust the grid resolution and change the grid size through the shrinkage and expansion of the grid,so as to capture arbitrary maneuvering target dynamics.For new targets randomly appearing in the tracking area,the sensor measurement analysis can be used to identify and generate new target grids for them.The experimental results show that the algorithm can realize the tracking of multiple maneuvering extended targets with unknown motion model and unknown intensity of new targets.(3)In view of the problem that the traditional tracking algorithm based on the ET-PHD filtering framework cannot estimate the complete trajectory of the target,the prediction correlation technology is introduced into the adaptive parameter particle PHD and adaptive grid-driven PHD algorithm for multiple maneuvering extended target tracking proposed in this paper.It is possible to carry out trajectory management on the estimation results of the filter,find the best correlation between the target state sets at adjacent moments,and give the continuous trajectory information of each target.The experimental results show the effectiveness of the method in estimating multiple maneuvering extended target trajectories.