一种基于小波分析的AR噪声滤波方法

摘要

为了滤除外测数据中AR模型的噪声，提出了一种基于小波分析的滤波方法。针对信号和AR噪声在小波域的分布特性，对传统的阈值滤波方法进行了改进，对小波系数进行偏自相关系数截尾检验来确定分解层数，利用广义交叉确认算法来确定分阈值，仿真结果表明该方法能有效滤除外测数据中的AR噪声。该方法对有色噪声的处理具有较大的借鉴意义。%In order to filter the noise of telemetered data in the AR model, a new filtering algorithm based on wavelet analysis is put out in the paper. By analyzing the distribution features of signal and noise in wavelet dimension, the traditional threshold filtering algorithm is ameliorated,using the partial autocorrelation coefficients truncated inspection for the wavelet coefficients to confirm decomposition level, using Generalized Cross Validation algorithm to confirm threshold. The result indicates that the new filtering algorithm can reduce the AR noise effectively Especially,the new filtering algorithm having importance significance for filtering the colored noise.