【摘要】 经典的ARMIA模型应用是对水文过程年际月变化所形成的时序数据进行计算处理,而忽略了水文过程平稳性检验和月际年变化对时序预测结果的影响.本文在对这一问题讨论的基础上,基于聚类提取分类后月份的特征,利用回归分析建立特征量和月水文数据间的关系,通过差分对特征量时序做平稳性处理,使用ARIMA模型按类预测特征量,由此,提出了一种新的挖掘水文时序月际年变化信息的方法,建立了改进的ARIMA模型及预测方法.作者以兰州降水站为例进行了应用验证,研究结果表明,改进后的ARIMA模型的精度要明显高于季节ARIMA模型,其平均残差达到了9.41,预报精度提高了21%,效果十分明显.最后就改进后的ARIMA模型的应用给出了进一步的研究方向.更多还原

【Abstract】 The ARMIA model is often used for calculating time series data formed by interannual variation with a month as unit.However,the influence brought about by inter-monthly variation with a year as unit is neglected.Based on the monthly data classified by cluster,the characteristics are extracted.The correlation between characteristic quantity and monthly data with a year as unit is constructed by regression analysis.To apply stationary treatment for characteristic quantity time series by difference,the ARMIA model is adopted for predicating characteristic quantity according to class.Therefore,a new method of data mining is put forward.An improved ARMIA model is developed and used for hydrological predication.The model is applied in Lanzhou precipitation station,and the result shows that the precision of the improved model is significantly higher than the seasonal model,the mean residual achieves 9 41,and the forecast precision is increased by 21%.Finally,some discussions about the application of the improved model are given.更多还原