基于聚类模糊加权马尔可夫链的降水量预测研究

Precipitation prediction based on clustering fuzzy weighted Markov chain

降水具有随机性,文章在加权马尔可夫链的基础上应用改进的ward系统聚类法对降水量进行聚类,引入优化后的隶属度对参照样本的状态向量进行预测,建立了聚类模糊加权的马尔可夫链模型,并取得了精度较高的预测结果。以北京市1963—2016年的年降水量作为训练样本进行模型训练,对北京市2017—2019年降水量进行预测,结果表明这3年降水量的预测值与真实值的相对误差均在3%以内,获得了较高的精度,能够为北京市水资源的合理规划提供依据。

The precipitation process has randomness.Based on the weighted Markov chain principle,this paper applies the improved ward system clustering method to cluster the precipitation,introduces the optimized membership degree to predict the state vector of the reference samples,establishes the clustering fuzzy weighted Markov chain model,and obtains the prediction results with high accuracy.Based on the annual precipitation data of Beijing from 1963 to 2016 as the training samples,the model training was conducted to predict the precipitation of Beijing from 2017 to 2019.The results show that the relative error between the predicted value and the true value of the precipitation in these three years is within 3%,and a high precision was obtained.It can provide a basis for the rational planning of water resources in Beijing.