量测提升策略下两级更新RBPF纯方位被动跟踪

Bearings-only passive tracking based on two stages update Rao-Blackwellised particle filter in lifting scheme of observation

针对采样型非线性滤波器在纯方位被动跟踪有效实现问题,提出了一种量测提升策略下两级更新RBPF算法。在标准RBPF算法框架下,通过对时间更新和量测更新过程的优化,以降低状态线性分量估计实现中的计算复杂度,给出了RBPF实现两级更新结构。另外,针对量测噪声随机性对于粒子权重度量的可靠性和稳定性的不利影响,结合系统建模中量测传感器精度信息实现虚拟量测集构建,并利用虚拟量测评估粒子重要性权重,通过减小粒子重要性权重方差,提高了重要性权重的估计精度。基于单站纯方位目标跟踪仿真场景,分析了算法滤波精度和计算量。理论分析和仿真实验结果验证了算法的可行性和有效性。

Aiming at the effective utilization of sampling nonlinear filter for bearings - only passive tracking, a no- vel two stages Rao - Blaekwellised particle filtering algorithm based on the lifting scheme of observation was pro- posed. In the framework of standard RBPF, the computational complexity from the estimation of state linear compo- nent was decreased by optimizing the realization steps of time update and observation update, and two stages up- date construction of RBPF was given. In addition, in order to improve the adverse effects on the measuring stability and reliability of importance weights caused by the randomness of observation noise, the virtual observation set was constructed by the information of sensor accuracy, and the importance weights was measured by those virtual obser- vations on this basis. Relative to importance weights measured by the current observation, the new method de- creased the variance of importance weights and improve the estimation precision of importance weights. The filte- ring precision and calculated amount of new algorithm was analyzed on the basis of single station bearings - only passive tracking simulation scene. The theoretical analysis and experimental results show the feasibility and effi- ciency of algorithm proposed.