Abstract: A new particle filter is proposed for the on-line estimation problem of non-Gauss nonlinear systems. In order to weaken the effect of historical information and enhance the effect of up-to-date measurement, it introduces attenuation memory factor for generating the important density function based on the Unscented Kalman filter（UKF） for a better performance in inhibiting the particle degradation problems in the new algorithm. As a result, the theoretical analysis and experimental results show that the new particle filter outperforms obviously superior to the standard particle filter and Unscented particle filter.

Key words: state estimation, particle filter, attenuation memory factor, important density function

摘要： 针对非线性、非高斯系统状态的在线估计问题，提出了一种改进的粒子滤波算法。该算法采用Unscented卡尔曼滤波器（UKF）产生系统的状态估计，并在量测更新过程中加入衰减记忆因子，消弱滤波器对历史信息的依赖，增强当前量测信息对滤波器的修正作用，从而产生一个优选的建议分布函数，较好地抑制了粒子退化问题。理论分析和实验表明：引入记忆衰减因子的粒子滤波，即衰减记忆无味粒子滤波（MAUPF）的性能明显优于标准的粒子滤波以及Unscented粒子滤波。

关键词: 状态估计, 粒子滤波器, 记忆衰减因子, 重要性概率密度函数