球面单形-径向容积粒子滤波的单站无源定位算法  

Single Observer Passive Location Based on Spherical Simplex-Radial Cubature Particle Filter

作　　者：张智[1] 姜秋喜[1] 潘继飞[1] 

机构地区：[1]解放军电子工程学院,安徽合肥230037

出　　处：《探测与控制学报》2014年第4期87-91,96,共6页Journal of Detection & Control

摘　　要：针对单站无源定位可观测性弱、观测噪声大而导致的定位精度低、收敛速度慢等问题,提出了球面单形-径向容积粒子滤波(Cubature Particle Filter,CPF)的单站无源定位算法。该算法基于容积卡尔曼滤波产生重要性密度函数,充分利用最新的观测信息,将粒子导向高似然区域。同时,其预测分布得到修正后的权值,有效缓解了粒子退化问题,提高了对系统状态后验概率的逼近程度。仿真结果表明,新算法虽然较标准的CPF增加了一定的计算量,但计算时间仍仅约为求积分粒子滤波(Quadrature Particle Filter,QPF)的10%,且定位精度与QPF相当,优于标准的CPF。Because of the low observability and the high measurement noise in single observer passive location, the performance of the location accuracy and convergence velocity was poor. A single observer passive location algorithm based on spherical simplex-radial cubature particle filter was proposed. The importance density function was obtained based on the cubature Kalman filter. The latest observation was utilized to generate the particles in the high likelihood area. Meanwhile, the modified weight was evaluated instead by the predictive distribution. The effect of the degeneracy problem was reduced, and the approximation to the system posterior density was improved. Simulation results indicated that the novel algorithm consumed about 10% as the computing time required by the quadrature particle filter, while the computation was a little larger than the standard cubature particle filter. Moreover, its location accuracy was comparable with the quadrature particle filter, and was higher than the standard cubature particle filter.

关 键 词：单站无源定位 容积粒子滤波 球面单形-径向规则 重要性密度函数 权值修正 

分 类 号：TN958.97[电子电信—信号与信息处理]