Methods Multi-sensor Target Tracking Passive State Estimation

Target tracking technology maintains the target current state estimation through processing target observation data, which has been widely applied for military and civil use. With the constant development and application of electromagnetic stealth technology and anti-radiation missile in modern electronic battlefield, passive multi-sensor target tracking technology has generally become as the hot research orientation by virtue of its good concealment and strong system survival capability. Based on the engineering practice, the paper studies the nonlinear estimation problem of passive multi-sensor target tracking technology. It offers the Gaussian Sum Particle Filtering method for passive multi-sensor single target tracking and Gaussian Sum Particle Probability Hypothesis Density Filtering method for passive multi-sensor multi-target tracking, which can provide beneficial reference for engineering practice of passive multi-sensor target tracking system.Considering the non-Gaussian distribution character of system estimated state and interfering noise in practical project, the paper put forward the Gaussian Mixture Particle Filtering method, which combines the Gaussian Mixture Model with common nonlinear estimation method to realize the passive multi-sensor single target tracking state estimation. Because system state is non-Gaussian distribution and system noise is Gaussian distribution, it regards the state distribution as Gaussian Mixture Model, combining with Extended Kalman Filtering to conduct single target tracing state estimation. Furthermore, as system state and system noise both are non-Gaussian distribution, it regards the state distribution, observation noise and process noise as Gaussian Mixture Model, combining with Particle Filtering method to conduct single target tracing state estimation. Simulation results demonstrate that Particle Filtering method under Gaussian Mixture Model featured with the similar estimated accuracy of Sequential Importance Sampling Resampling Particle Filtering and shorter time-consuming when passive multi-sensor target tracking possesses non-Gaussian distribution character.Considering the complex calculating of multi-target tracking data association process, the paper applies Probability Hypothesis Density Filtering method to realize passive multi-sensor multi-target tracking state estimation. According to Random Finite Sets theory, it modeling the respective collections of targets and measurements as random finite sets and the PHD of them as the sum of multi weighting Gaussian components. Then it conducts the prediction and updating combining with Monte Carol and particle filtering method, which can avoid data association in traditional multi-target tracking and realize multi-target tracking state estimation. The simultaneous result indicates that this method is as effective as Joint Probabilistic Data Association and Sequential Importance Sampling Resampling Particle Filtering for passive multi-sensor multi-target tracking state estimating. But this method’s applicable range, it can be applied to the actual engineering environment with the uncertainty target number.