融合粒子滤波与交互多模的单站被动定位算法

Mono-station passive location algorithm based on particle filter and interacting multiple model

融合交互式多模型和粒子滤波算法提出了一种新的单站被动定位算法。该算法采用多模型结构,各模型匹配粒子滤波算法,从而适用于非高斯系统噪声、非线性量测方程条件下,具有任意运动轨迹的目标跟踪问题。各模型粒子经过输入交互,充分体现交互式多模型优点。粒子滤波的重要密度函数采用UKF方法产生,使得较少的粒子即可更加接近系统状态的真实后验概率,从而减少了运算量。仿真结果表明:本文新算法性能明显优于标准交互式多模型算法,具有一定的工程实用价值。

Combining particle filter with interacting multiple model （ IMM）, a novel mono-station passive location algorithm is proposed. The algorithm uses the multiple models and every model matches particle filter in order to track arbitrary maneuvering target with non-Gaussian system noise and non-linear measurement equation. Particles in every model are interacted firstly to show the advantage of the IMM algorithm. The particles are more approximates to the posterior density of the states by using UKF method to generate the importance density function, so the computation complexity is reduced. Simulation results show that the proposed algorithm is more efficieney than general IMM algorithm and has certain practical utihty for engineering application.