交叉熵相似性度量在水文时间序列匹配中的应用

摘要

Similarity search algorithm based on cross entropy uses wavelet transform for data preprocessing and use PAA to reduce dimensionality .The approximate probability distribution of the water level data is obtained .By cross‐entropy distance ,the probability distribution of the difference between two sequences are measured .Taking Kang Hill station of Poy‐ang Lake water level data as an example ,this algorithm verifies the cross‐entropy‐based similarity search algorithm to accu‐rately identify a given period of time compared to a similar hydrological processes under the premise of reducing the time com‐plexity .%基于交叉熵的相似度查询算法利用小波变换对数据进行预处理，通过 PAA 算法对数据进行降维，求得降维后的水位数据的近似概率分布情况。通过交叉熵距离来度量两个序列间的概率分布差异。以鄱阳湖星子站日水位数据为例，验证了基于交叉熵的相似度查询算法在降低时间复杂度的前提下较准确地找出给定时间段相似的水文过程。