中文摘要：

由于传统的增广卡尔曼滤波方法难以有效解决带有未知参数估计的强非线性、非高斯动力学问题。针对这一问题,在对粒子滤波算法研究的基础上提出了基于近似思想的增广粒子滤波方法。这一方法利用高斯随机游走模型对未知参数进行增广建模,再通过粒子滤波方法进行状态估计。为了提高观测新息的利用率,提出了一种新的重要性函数;针对高斯随机游走模型方差不断增大的问题,采用了修改后的Kernel平滑模型进行解决;对粒子重采样方法进行了修改,采用了混合重采样的策略,增强了粒子活性。通过算例进行仿真,验证了算法的有效性。

英文摘要：

Because the tradition Kalman Filter could not solve the problem of unknown parameters estimation in nonlinear and/or non-Gaussian dynamic system, aiming to this problem, a new algorithm was proposed.＂ Extension Particle Filtering （PF） based on the Thought of Approximately, which was based on the study of PF algorithm. This algorithm used Gaussian random walk process to model unknown parameters, and then estimated the state variation by particle filtering algorithm. In order to improve the observing information effectively, a new important density was proposed. In order to solve the problem that covariance augmented infinitely with time in Gaussian random walk model the Kernel smooth model was modified. Then, a mix resampling method was proposed to improve the active of particles. Finally, the effectiveness of purposed algorithm was validated by an example.