基于灰色马尔可夫链的径流序列模式挖掘

Runoff series pattern mining based on grey Markov chain

河川径流受气候、地貌、土壤、植被等自然条件以及人类活动的耦合作用,其演变过程既表现出灰色禀性,同时也有强烈的随机性.探讨了径流序列模式挖掘的灰色系统方法,提出了综合考虑径流序列趋势变化与随机变化的灰色马尔可夫预测方法,并用于年径流序列预测.通过黄河上游贵德站年径流序列资料验证,该方法比GM(1,1)有更高的精度,满足规范要求,计算方法可行.结果可为黄河上游防洪抗旱、区域水资源管理、水利水电工程规划提供科学的依据.

River runoff is affected by coupling action of human activity and natural conditions, including climatic change, topographical features, soil and vegetation cover. Its change process shows both grey character and strong randomness. In this paper, grey system model of runoff series pattern mining is discussed, considered trend change and stochastic change of the runoff series; and forecasting method based on grey Markov chain is presented to predict annual runoff. Example of Guide station in the upper reaches of the Yellow River proved that forecast precision is higher than GM （1,1） ; the results meet the current standard; and this method is feasible. The conclusion can provide scientific base for flood controlling, drought resisting, water resource management and hydroelectric engineering planning in the upper reaches of the Yellow River.