基于生命旋回-权马尔可夫的径流预测模型

A prediction model for river annual runoff based on life cycle-weighted Markov

针对生命旋回预测模型不能反映河川径流周期性和波动性变化的特点,将其与权马尔可夫链结合,提出了一种精度较高的预报模型——生命旋回-权马尔可夫链组合模型.该模型用生命旋回模型预报河川径流的趋势项变化,由权马尔可夫链对径流残差序列进行修正,进行预测时采用等维信息方法处理,在此基础上对黄河龙门水文站径流进行预测,拟合精度为86.97%,合格率为86.78%,表明该模型可以用于径流预测.此外,对求解生命旋回模型的方法作出了改进,经研究表明,该方法实用可行.

The river annual runoff prediction is very important to hydraulic and hydropower engineering management; and it is also a complex problem. The life cycle model is a new long term prediction model for river runoff, which requires less data, simple calculation. However, due to the physical mechanism, the prediction results of the life cycle model can＇t reflect the fluctuate and random characteristics of the river annual runoff; and it can not meet required precision for hydraulic and hydropower engineering management, which usually need one year prediction period. Aiming at these problems, the life cycleweighted Markov combination prediction model is put forward, in which the life cycle model is used to predict the tendency item of the river annual runoff; and the weighted Markov model is also involved to predict the random item. The annual runoff at Longmen Station of the Yellow River from 1957 to 2001 is predicted with the new combination model; and the annual runoff from 2002 to 2005 is tested. The resuits show that： the average precision of the combination model is 86.97%; 86.78% prediction results is qualified. In addition, this paper improved the seeking method for parameters of the life cycle model, and it is feasible.