基于经验模态分解的年径流自忆性预测模型  

Self-memory Model Based on Empirical Mode Decomposition for Annual Runoff Prediction

作　　者：袁潇晨[1] 金菊良[1] 王宗志[2] 张明[3] 

机构地区：[1]合肥工业大学土木与水利工程学院,安徽合肥230009 [2]南京水利科学研究院水文水资源与水利工程科学国家重点实验室,江苏南京210029 [3]安徽工程科技学院生化系,安徽芜湖241000

出　　处：《水电能源科学》2010年第6期1-3,169,共4页Water Resources and Power

基　　金：国家自然科学基金资助项目(70771035;50909063);中国气象局成都高原气象开放实验室基金资助项目(LPM2008018)

摘　　要：针对径流量时间序列非线性、非平稳性的特点,利用经验模态分解法将其分解为多个不同频率下的时间序列组合揭示演变规律,采用双向差分法对分解后的各序列逐个反导其微分方程,并结合自忆性原理构建预测模型,由叠加分解后的各序列预测值获得径流量的预测值。实例结果表明,该方法充分挖掘了数据自身信息,拟合效果较好,预测年径流量准确,具有推广应用价值。Runoff time series have characteristics of non-linear and non-stationary.In order to reveal the evolution law of runoff,empirical mode decomposition method which is self-adaptive is used to decompose them into some time series with different frequencies.Taking a differential equation which is retrieved from one of the decomposed time series based on the bilateral difference principle as a dynamic kernel,the self-memorization forecast model is established.Each of the decomposed time series can get a value fr...

关 键 词：经验模态分解 径流预测 自忆性 加速遗传算法 

分 类 号：TV121[水利工程—水文学及水资源]