基于无味有向图切换的机动目标跟踪VSMM算法

A VSMM Algorithm Based on Unscented Digraph Switching for Maneuvering Target Tracking

在机动目标跟踪领域,传统的基于固定模型集合的交互式多模型算法需要大量的模型来保证模型不失配,但是会带来巨大的计算量,并且来自过多模型的不必要的模型间竞争反而会使算法性能下降。为解决此矛盾,该文提出了一种基于有向图切换和无味卡尔曼滤波算法的变结构多模型算法。该算法根据目标不同时刻的运动状态依据有向图匹配不同的子模型集合进行滤波。仿真结果显示该算法在有效降低计算量的同时使模型集合和目标的运动状态更好的匹配,从而提高了算法跟踪精度。

In the region of maneuvering target tracking, the traditional interacting multiple model algorithm has a fixed structure, which brings unnecessary inter-competition among the amount of models and calculation burden that the algorithm needs enough models to ensure the accuracy. To resolve this contradiction, a variable structure multiple-model algorithm based on digraph switching and unscented Kalman filter is presented. According to the target motion state basis at different times, the algorithm can match sub-model-set based on the digraph. Simulation results show that this algorithm improves the accuracy of tracking by reducing the competition of the models as well as reducing the computation burden.