处理有色观测噪声的粒子滤波算法

Particle Filter for Colored Measurement Noise

针对经典Kalman滤波无法直接处理有色噪声的问题,采用多项式长除法将有色观测噪声模型展开成无穷级数,截断取其有限项获得有色噪声的先验信息;然后利用粒子滤波能够处理非高斯噪声的特点对有色观测噪声进行处理。通过一个GPS定位算例,将此新方法与观测扩增方法进行了分析和比较。结果证明,利用该方法能有效地控制有色观测噪声的影响。

Polynomial-quotient has been used, aiming at solving problem of the colored measurement noises, which translates colored observation noises into infinite series, and the variances of colored observation noises have been calculated. Particle filter was followed to estimate the parameters. In order to verify the validity and rationality of this method, a contrast between this method and the approach of observation expand filter was given. The result showed that the the influences of the colored observation noises effectively could be controlled in this approach.