马氏链预测方法的统计试验研究  

Study on Statistical Experiment of Markov Chain Prediction Methods

作　　者：夏乐天[1] 朱元甡[2] 

机构地区：[1]河海大学理学院,南京210098 [2]河海大学水资源环境学院,南京210098

出　　处：《工程数学学报》2010年第2期313-320,共8页Chinese Journal of Engineering Mathematics

基　　金：河海大学自然科学基金(2084408319)~~

摘　　要：近年来,基于绝对分布的马尔可夫链预测(ADMCP)方法、叠加马尔可夫链预测(SPMCP)方法和加权马尔可夫链预测(WMCP)方法在各种水文序列的预测理论中得到了广泛的应用。尽管各种马尔可夫链预测方法在有关应用实例中效果都不错,但是,实际上它们的预测效果是有差别的。本文通过统计试验手段对这三种马尔可夫链预测方法进行了比较研究,结论表明:WMCP方法预测精度最高,SPMCP方法预测精度次之,ADMCP方法预测精度相对来讲最差。In recent years, the weighted Markov chain prediction method （WMCP）, the Markov chain prediction method based on absolute distribution （ADMCP） and probability summation （SPMCP） have been widely used in the various hydrology series. Although the application results of various Markov chain prediction methods in real situations are seemed good, the prediction accuracy of these three kinds of Markov chain prediction methods is different. The three prediction methods are studied by using statistical experiment. The experiment results indicate that the prediction accuracy of the WMCP method is the highest, the ADMCP method is the lowest, and the SPMCP method is in the middle.

关 键 词：马尔可夫链 水文序列 预测 统计试验 

分 类 号：P334[天文地球—水文科学]