基于灰色马尔柯夫预测模型的径流量预测

Prediction of Runoff Based on Grey Markov Model

针对河川径流成因复杂性的特点和用单一预测法均有一定局限性的现状,提出了灰色与马尔柯夫相耦合的灰色马尔柯夫预测模型。两种预测模型的科学组合,既综合了GM(1,1)灰色预测和马尔柯夫预测的优点,又提高了预测流域径流量的精度。讨论了GM(1,1)模型修正法和相对误差序列的"马氏性"检验法,进一步完善了该预测模型。最后以安康水库年入库径流量预测为例,验证该方法的可行性。结果表明:1995年和1996年入库径流量的预测值分别是以0.41,0.39的最大概率落入区间(99.894,139.592)和(101.088,142.509)内,由此可见,预测结果准确。

Aiming at the complexity of runoff cause and limitation of a single prediction method applied,a new method,called grey Markov model,is presented based on analyzing grey theory and Markov theory.The two prediction methods were scientifically combined,which generalizes advantages of the ones,raises the accuracy of runoff prediction.Furthermore,the methods of GM（1,1） model updating and Markov property testing on series of relative error were also discussed.So the model was further improved.Finally,this prediction model was identified by taking prediction of Annual Runoff Variation into Ankang Reservoir.The results showed that predicted value of 1995,with maximum probability 0.41,falled in the interval（99.894,139.592）and predicted value of 1996,with maximum probability 0.39,falled in the interval（101.088,142.509）.Thus,the predicted result was accurate.