基于加权马尔可夫模型的径流预测研究

Application of Weighted Markov Chain Model to Predict Runoff

本文尝试应用模糊C均值聚类(FCM)划分马尔可夫链模型的径流状态区间,再针对径流量随机变量特点,以各阶自相关系数为状态权重,应用该加权马尔可夫链模型预测径流量状态。依据青海省达日县吉迈水文站55年(1950~2004)年径流量资料,预测了2001~2004年的径流状态,预测结果均与实际情况相符合。说明用FCM确定状态区间的马尔可夫链模型对吉迈水文站的径流状态预测是可行的、有效的。

This paper tries to take the Fuzzy C-Means method (FCM) to classify runoff state of Markov chain model. Then aiming at the runoff characteristics of dependent random variables and taking order of autocor- relation coefficients as the weights, Markov chain model is used to predict the following year runoff state. The runoff state from 2001-2004 are predict based on the 54 years annual runoff data from 1950-2004 at Ji- mai gauging station located in Dari country, Qinghai Province, the results are consistent with the actual situa- tion. That means the application of Markov chain model with FCM to determine runoff state of Jimai gauging station is feasible and effective.