中文摘要：

在小波多尺度分解的基础上,首先利用小波谱识别出所研究时间序列的主尺度,然后采用正弦拟合对主尺度的主频信息进行消除,以达到识别被主频信息掩盖的异常信息的目的。通过对湖南省安乡井水位日均值序列和河北省浅牛6井水位日均值序列的分析处理实验,获得了较好的结果。

英文摘要：

Basing on the wavelet multi-scaling decomposition, we make use of wavelet spectrum to identify the main scale of the studied time sequence, and then eliminate the main frequency information of the main scale by sine function fitting to identify water level anomaly covered by main frequency information. We analyze daily mean value sequences of water level of Anxiang well in Hunan and Qianniu well in Hebei by wavelet multi-scaling decomposition. The results show that the method is useful to identify water level abnormal information covered by dominant frequency information.