基于FARIMA模型的日流量过程随机模拟研究  

Stochastic simulation of daily river flows based on FARIMA model

作　　者：郭辉[1] 齐梅兰[1] 刘春光[1] 魏金[1] 

机构地区：[1]北京交通大学土木建筑工程学院,北京100044

出　　处：《水力发电学报》2012年第6期22-27,共6页Journal of Hydroelectric Engineering

基　　金：国家自然科学基金(51078021);教育部高等学校博士学科点专项科研基金(20100009110016)

摘　　要：本文采用能反映时间序列长相关性的自回归分数整合滑动平均(FARIMA)模型,建立了日流量过程的随机模拟方法。首先提出概率权重矩-正态分位数变换(PWM-NQT)法进行日流量预处理,将预处理后得到的平稳、正态时间序列用于FARIMA模型的辨识和参数估计。基于FARIMA模型得到正态化日流量模拟值,并由正态分位数逆变换及逆标准化分别得到标准化日流量模拟值及日流量过程模拟值。将建立的模型应用于长江流域宜昌水文站日流量过程随机模拟,统计检验表明,建立的随机模拟方法能合理考虑日流量过程的季节性周期和长相关性,同时也能很好地保持日流量过程的其它统计特性。A fractal auto-regressive integrated moving average（FARIMA） model suitable for time series of long-term persistence is proposed and used to develop a stochastic model of daily river flows.First,stationary and normalized daily river flows are obtained from the observed series with a probability-weighted moment method coupled with normal quantile transform（PWM-NQT）,and these flows are then used for identification and estimation of the FARIMA model and its parameters.Synthetic series are generated by this model and transformed into standardized series and simulated series by inverse of NQT and inverse of standardization,respectively.A case study of the model was made for simulation of the daily flows at the Yichang station on the Yangtze.The validation results show that the proposed method is proper for consideration of seasonal periodicity,long-term persistence and other major statistical properties of daily river flows.

关 键 词：水文学 随机模拟 FARIMA模型 日流量过程 PWM-NQT 实用性检验 

分 类 号：TV121.4[水利工程—水文学及水资源]