Mixed-Effects Model And Its Application Of Multiple Targets Tracking

Multiple targets tracking research has been focused strongly with the rapid development of the modern information war and the traffic control of air and sea.A standard assumption of most multiple targets tracking algorithm is that the targets' movement were independent of each other with known parameters.It can be discussed by building state space models.However,in fact,multiple targets always move as a group,and the parameters of the individual are usually unknown.Based on the above situation,this paper extends state space model to mixed-effects state space model,and explores the method of modeling and algorithm of state estimation by combing mixed-effects state space model with multiple targets tracking problem.Firstly,based on complex data association problem,we adopt the improved multiple hypothesis tracking(MHT)algorithm to explore methods to the establishment of the linear Gaussian model with known parameters,and extend it to the nonlinear model with unknown parameters,and then build mixed-effects state space model-multiple targets tracking.Secondly,the algorithm of state estimation of multiple target tracking models with known parameters has been discussed.The Gaussian mixture model is used for recording multiple hypotheses,and measurement information is used for solving the uncertainty of data association,then state estimation is obtained as a result.Finally,to solve state estimation problem of mixed-effects state space model-multiple target tracking,an algorithm combining particle filters algorithm based on sampling thought with Kalman filter is proposed,and through the improvement,the algorithm of state estimation combining auxiliary particle filters based on kernel smoothing with Kalman filter is obtained.In view of the proposed algorithm,in this paper a ground target tracking model with two random disturbances has been taken for numerical simulation.By taking the established models based on particle filters algorithm and its improvement of auxiliary particle filters algorithm for state estimation,the results indicate that auxiliary particle filter algorithm based on kernel smoothing is a priority selection.Further,the consistency check on the preferred algorithm also gets accurate results.Thus the algorithm proposed in this paper by combing mixed-effects state space model with multiple targets tracking problem for modeling and state estimation has important application values.