排序方式: 共有3条查询结果,搜索用时 0 毫秒
1
1.
2.
提出了一种改进的不敏粒子滤波(UPF,Unscented Particle Filter)算法。和传统的UPF相比,该算法有两点改进,首先,在形成"粒子云"时,直接采用当前时刻各粒子的UKF(Unscented Kalman Filter)估计作为粒子,在保证粒子有效性的同时,减少了UKF之后的重采样过程;然后,结合新的粒子产生办法,重新定义了权值计算方法,避免了对各粒子重要概率的复杂计算。仿真表明,改进算法在减少计算量的同时,有效地提高了跟踪稳定性和跟踪精度。Abstract: An improved Unscented Particle Filter (UPF) algorithm was proposed.Compared with traditional UPF,it has been improved at two points.First,when producing particle cloudy,it directly uses the current particle estimation of Unscented Kalman Filter (UKF) as new particle,which guarantees the validity of particles and eliminates the re-sampling process after UKF as well;then,according to the new particle-producing method,a weight-calculating formula is re-defined,so as to avoid the complicated computation of proposal probability of every particle.A simulation shows that the Improved UPF (I-UPF) can effectively enhance the tracking stability and tracking precision and reduce computational cost at the same time. 相似文献
3.
一种姿态角辅助的IMMPF算法 总被引:1,自引:0,他引:1
交互式多模型粒子滤波(IMMPF)算法把粒子滤波(PF,Particle Filter)引入交互式多模型估计(IMM)算法,从而能够有效地解决非线性、非高斯机动目标跟踪问题.首次把姿态角信息引入到针对空中机动目标跟踪的IMMPF算法中,通过姿态角测量与当前运动模式的模糊关联来辨识目标的机动模式,然后把辨识结果与IMMPF算法的后验粒子权值相融合,以提高算法本身的模型分辨能力;对机动目标跟踪的仿真实验表明,该方法能够有效地改善原跟踪算法的跟踪精度和稳定性. 相似文献
1