基于扩展符号聚集近似的水文时间序列异常挖掘

摘要

水文时间序列异常挖掘目前大多采用基于距离的方法.为了克服该方法耗时长、计算量大的缺点,采用一种符号化算法,用扩展符号聚集近似对序列符号化表示,再对字符串进行距离度量,并以太湖流域小梅口站逐日水位数据为例进行验证.实验表明该方法的挖掘结果更全面,运算效率很高,更适合处理大规模数据集.%Most of anomaly mining of hydrological time series uses distance based method. Since the method was time-consuming and has a great deal of computation, this paper applied extended symbolic aggregate approximation and then measured distance of strings. It verified the validity of the method by the water level data obtained from Xiaomeikou gauge station in the Taihu Lake. The experimental results show that the method has high efficiency and is more suitable for processing large-scale data sets.