复合K噪声下机动目标跟踪自适应UPF算法

摘要

Aimed at the strong nonlinear and non-Gaussian characteristics of maneuvering target tracking system under compound K noise, an adaptive unscented particle filter (AUPF) algorithm is proposed. Based on constant acceleration (CA) model and its modified filtering algorithm, the algorithm adopts a new proposal distribution which combines unscented Kalman filter (UKF) and strong tracking filter (STF) and enhances the system performance for tracking general mobile and step mobile. The AUPF algorithm is applied to track several kinds of typical maneuvering targets based on the model of compound K noise. And the comparison with the unscented particle filter (UPF) algorithm is given. The simulation results show that AUPF algorithm has good track performance for tracking various maneuvering targets and has high tracking precision.%针对复合K噪声下机动目标跟踪系统具有强非线性非高斯的特点,提出了一种自适应无迹粒子滤波(Adaptive Unscented Particle Filter,AUPF)算法.该算法建立在常加速模型及其改进滤波算法基础上,并将无迹卡尔曼滤波(Unscented Kalman Filter,UKF)与强跟踪滤波(Strong Tracking Filter,STF)算法相结合作为提议分布,提高了系统跟踪一般机动和阶跃机动的能力.在给出复合K噪声模型的基础上,利用AUPF算法对几种典型机动目标进行了计算机仿真,并同无迹粒子滤波(Unscented Particle Filter,UPF)算法进行了比较.仿真结果表明,复合K噪声下AUPF算法能更有效地对各种机动目标进行跟踪,具有较高的跟踪精度.