基于自适应变结构信息滤波的目标跟踪算法

目前目标跟踪算法采用的交互多模型，大多是通过固定模型之间的切换来完成目标跟踪，这容易出现模型集与目标真实运动不匹配问题，降低目标跟踪的精度。同时，现在大部分观测平台都能提供多传感器量测，这要求跟踪算法能对不同量测信息进行高效数据融合。针对上述问题，提出一种基于自适应变结构多模型和信息滤波的跟踪算法，它由少量模型构成模型集，通过在线更新模型集参数以自适应目标真实运动，采用无迹卡尔曼信息滤波融合多传感器量测信息，实现对目标的跟踪。仿真结果表明，该算法可以有效融合多传感器量测信息，自适应匹配目标真实运动，实现对目标稳定的高精度跟踪。

New maneuvering target tracking algorithm based on adaptive variable structure multi-model and information filtering

Most current maneuvering target tracking algorithms use Interacting Multiple Model （IMM） by switching fixed models, which can easily cause problems like model-mismatching and thus resulting in inaccurate tracking. Meanwhile, multiple sensors are now generally available to provide multi-measurements for target tracking, provoking the needs to efficiently utilize them. To solve these problems, a new algorithm that consists of less models and integrates adaptive Variable Structure Multi-Model （VSMM） and Information Filtering （IF） is presented, which accomplishes better estimation of maneuvering target’s trajectory through updating the model-set’s parameters in real time and fusing multi-measurements by unscented Kalman IF. Simulation results prove that this new algorithm can efficiently fuse multi-measurements, adaptively match target’s real model and reach a high level of target tracking.