基于灰色预测的粒子滤波算法

A particle filter algorithm based on gray forecast

针对标准粒子滤波（SPF）算法在目标机动时跟踪滤波性能不高的问题，引入灰色预测理论，提出了一种基于灰色预测的粒子滤波（PFGF）算法，给出了算法的具体描述和运算流程。当预先建立的状态模型不再适用于目标的真实运动状态时，该方法具有良好的预估性能，减少了对事先假定目标状态模型的依赖性。与SPF方法进行了蒙特卡洛仿真比较分析，实验结果证明，PFGF算法在不增加计算复杂度的情况下，提高了跟踪精度，能够很好地克服粒子退化现象。

The standard particle filter （SPF） algorithm＇ s problem of low targer-tracking performance due to the serious granule degeneration that occurs when the state model deviates the target＇ s motion state is paid attention, and aiming at this, a particle filter algorithm based on the gray forecast theory, called the PFGF algorithm is presented with the detailed description of it. When the condition model established in advance is no longer suitable for the goal＇ s proper motion condition, this new algorithm is with the good estimate performance. It reduces the dependence on the beforehand target condition model. The Monte-Carlo simulation results show that the new algorithm increases the tracking accuracy without increasing the computation complexity compared with the SPF algorithm. It can overcome the phenomenon of granule degeneration effectively.