Research Of Jerk Model Tracking Algorithm On Highly Maneuvering Target

Target tracking technology, a science research method of a series of estimating andforecasting calculation of the target’s current and future motion state, according to the targetinformation about position or course angle and the others obtained by various sensors. It iswidely used in many fields no matter on military or civil products or many other areas. Itsmain research contents are trying its best to estimating and forecasting the current realinformation of the targets from the imprecise target information accurately. In this process,doing some filter work toward those collected data information is absolutely necessary.After decades of development, Kalman filter has been proved to a good filter algorithmwhich has excellent performance. At the meanwhile, in general, the filter process is on thebasis of certain target tracking models, so, we summarize that maneuvering target trackingtheory includes two main research objects: how to model the target’s movement locus andchoose the right filtering algorithm. Science and technology has a high level increasingnowadays, with the increasing maneuver performance of motor vehicle, referencingpredecessors’ researches, toward those highly maneuvering targets, we do the followingacademic researches work in this paper:Firstly, summarize development status of these years and research status both in homeand abroad of maneuvering target tracking. Introduce measurement data, modelingmaneuvering targets, adaptive filter and forecast of maneuvering trajectory, as well asmaneuvering detection and maneuvering identification of the targets.Secondly, introduce two research objects of maneuvering target tracking area in detail:modeling and state estimation of the maneuvering target. Respectively, summarize thosebasic models such as CV model and CA model that are used for the non-maneuveringtargets. Those basic models such as Singer model, CT model, Jerk model and CS model thatare used for the maneuvering targets. Meanwhile, for those combination models, includeIMM model and CS-Jerk model, make some detailed analysis about their principles. As totargets state estimation, emphasis on some commonly used linear filter methods: KF, linearself-regression filter, filter, and two points extrapolation filter; non-linear filtermethods: PF, EKF, STF and UKF.Finally, through analyzing the CS-Jerk model, as to the problem of maneuveringfrequency is presetted by human in the model, put forward a much more reasonable modified maneuvering target tracking algorithm on the basis of CS-Jerk model. In thisalgorithm, firstly, design a fuzzy controller of adaptive realization of maneuvering frequency.And then, according to the measurement innovation and its change rate, this algorithmadjusts maneuvering frequency in the CS-Jerk model adaptively by fuzzy controllerdesigned before. And then, By analysing the principle and the characteristics of strongtracking filter，this paper we select STF as the filter algorithm because it has very nice filterperformance towards those high speed and highly maneuvering targets. Through MonteCarlo method, simulation the original algorithm and modified algorithm. Results show thatmodified maneuvering target tracking algorithm significantly increases trackingperformance comparison with the original algorithm, while time complexity is about thesame. Reaches the expected goal, has some good reference value for the engineeringapplication.