基于新的数值积分粒子滤波的机载无源定位算法  

Novel Cubature Kalman Particle Filter for Airborne Passive Location

作　　者：刘学[1,2] 焦淑红[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院 [2]中国人民解放军91245部队

出　　处：《宇航学报》2011年第7期1478-1485,共8页Journal of Astronautics

基　　金：973国家安全重大基础研究基金资助课题(61393010101-1);黑龙江省教育厅科学技术研究项目(11544025)

摘　　要：针对机载无源定位这一多维非线性滤波问题,提出一种新的用3阶数值积分卡尔曼滤波算法来产生重要性密度函数的粒子滤波算法。新算法采用球形和径向数值积分规则选取积分点和确定相应的权值,得出的积分点数仅为状态维数的二倍,大幅的减少了计算量,较好地解决了求积分卡尔曼粒子滤波算法(Quadrature KalmanParticle Filter,QPF)在高维滤波时存在计算量大的问题;而且通过设定比例因子使得所产生的重要性密度函数在系统状态转移概率密度的基础上,融入最新的观测值,增加了粒子的多样性,提高了对系统状态后验概率的逼近程度。仿真结果表明:新算法在稳定性和定位精度上与QPF相当,但计算时间仅约为QPF的15%。In order to solve the question of airborne passive location of multi-dimensional nonlinear filtering,a novel Particle filter algorithm is proposed by using three order Cubature Kalman filter to generate the importance density function.The new algorithm adopts Spherical rule and Radial rule to calculate the quadrature points and determine the corresponding weights with the final quadrature points only twice the state dimension,which considerably provide a systematic solution for the computational complexity problems encountered in the Quadrature Kalman filter under high-dimensional nonlinear filtering,and the new algorithm also sets scale factor to integrate the latest observations with the figured importance density function on the basis of the system state transition density,thus increasing diversity of particles,and improving the approximation to the system posterior density.Simulation results indicate that the Cubature Kalman Particle filter only consumes about 15% the computing time required by the Quadrature Kalman particle filter,while the stability and location accuracy is kept.

关 键 词：无源定位 粒子滤波 求积分卡尔曼滤波(QKF) 

分 类 号：TN957.51[电子电信—信号与信息处理]