基于滑动移除近似熵的泾河径流序列变异分析

摘要

The use of Moving Cut Data-Approximate Entropy made variation analysis for runoff series of Zhangjiashan Station on the Weihe River in the period of 1960—2001. The results show that the runoff series of Zhangjiashan Station have undergone abstract variation in the year of 1967,1971,1983 and 1996. Comparing with both Moving Approximate Entropy and Mann-Kendall methods,the result of Moving Cut Data-Approximate Entropy,which is depended less on moving window and which has higher detection accuracy,is stable,convenient and accuracy.%利用滑动移除近似熵对泾河张家山站1960—2001年径流序列的变异点进行了分析。结果表明：张家山站径流序列在1967年、1971年、1983年、1996年均发生了变异；相对于滑动近似熵和 Mann － Kendall检验法，滑动移除近似熵的诊断结果更加稳定，具有对滑动窗口的依赖性小、方法简便、检测准确性高等特点。