【摘要】 当观测数据中含有粗差(异常值)时,其误差分布可视为污染分布。卡尔曼滤波对异常值非常敏感,它严重影响动态定位的精度。本文应用Bayes定理,给出测量噪声为污染正态分布时的一种Bayes估计动态模型,通过模拟和实例计算与比较分析,表明算法的有效性。更多还原

【Abstract】 The observations are sometimes contaminated by gross errors (or outliers) and the error distribution in this case can be regarded as contaminated distribution. The precision of kinematic positioning by using Kalman filtering is very sensitive to the outliers. A Bayesian estimator is given for dynamic models with contaminated distribution, based on Bayesian theory. Simulation and practical examples of filtering are given, which show that the modified Kalman filtering is effective and reliable.更多还原