基于高斯粒子滤波的当前统计模型跟踪算法

对于非线性系统估计问题,高斯粒子滤波器可以获得近似最优解,与粒子滤波器相比其优点是不需要重采样步骤和不存在粒子退化现象.采用高斯粒子滤波代替当前模型自适应跟踪算法中的卡尔曼滤波,将高斯粒子滤波与当前统计模型的优点相结合,提出了一种新的当前统计模型自适应跟踪算法,用于非线性非高斯系统的机动目标跟踪.MonteCarlo仿真表明,该算法跟踪精度优于标准的交互多模型算法和当前统计模型自适应跟踪算法,实时性好于交互多模型粒子滤波算法.

Current statistical model tracking algorithm based on gaussian particle filtering

Gaussian Particle Filter (GPF) is asymptotically optimal for nonlinear system estimation problems. The advantage of GPF over the Particle Filter (PF) is that it does not need the re-sampling step and avoids the particle degeneracy phenomenon. A new current statistical model tracking algorithm is proposed, which is applied to maneuvering target tracking in non-linear and non-Gaussian system. In the new algorithm, replacing Kalman Filter with GPF in current statistical model adaptive tracking algorithm integrates the advantages of GPF with the ones of current statistical model. A simulation of a maneuvering target tracking model is presented. The simulation shows that the tracking accuracy of current statistical model adaptive tracking algorithm based on GPF is superior to that of standard interacting multiple model algorithm and current statistical model adaptive tracking algorithm, the tracking speed of which is better than that of Interacting Multiple Model Particle Filter algorithm.