基于改进Kalman滤波算法的粗差修正及应用

现代动态数据处理中，以Kalman滤波为代表的现代时间序列分析方法发挥着举足轻重的作用。对于平稳性比较好的数据，一般方法都能够得到良好的处理结果，并可以做出相对准确预测，但是在动态数据中总会出现一些粗差值或异常值，如果这些异常数据不能得到有效的处理控制的话，不仅会对本期处理结果造成严重影响，还会因计算过程中状态值吸收此项粗差而对后期数据的处理结果产生影响。针对可能出现的粗差项及其带来的影响，本文首先详细地讲解了Kalman滤波的具体理论，在此基础上提出了一种改进的基于Kalman滤波的修正方法，经过实际实验发现，在无需先探测粗差的情况下，本方法能够对所含粗差项进行有效的修正并且效果良好。

Gross Error Revision of Filter Algorithm and Its Application in Improved Kalman

In the field of modern dynamic data processing, time series analysis method is the most important one especially the Kalman Filter. If the data has good stability, all the normal methods can get the good result, even producing the much more exact forecast. But for the dynamic data, gross error is very common and can not be ignored. If those abnormal data can not be processed very well, it could affect the result of this time and be absorbed during processing that would affect the result behind this. For the gross error and its bad impact, we introduce an improved Kalman Filter algorithm and we apply this algorithm in dynamic data processing that has unknown gross error, and we can find out the gross error and correct it.