基于Holt-Winters方法的年径流序列周期识别  

Cycle Identification of Annual Runoff Time Series Based on Holt-Winters Method

作　　者：袁嘉晨 陈元芳[1] 顾圣华 许广东[1] 

机构地区：[1]河海大学水文水资源学院,江苏南京210098 [2]上海市水文总站,上海200232

出　　处：《水电能源科学》2016年第3期28-31,共4页Water Resources and Power

基　　金：水利部公益性行业科研专项经费项目(201201068;201301066);2013年中央高校基金(B12020118);城市化的洪灾风险效应及适应性风险管理体系研究(51479061)

摘　　要：针对小波分析法在年径流序列周期识别方面存在的不足,以长江支流乌江水系武隆站为例,利用Holt-Winters模型固有结构模拟水文时间序列,根据拟合精度来判断年径流序列周期出现的可能性,以达到周期识别的目的,并与小波分析法识别的序列周期进行比较。结果表明,Holt-Winters模型很好地捕捉到了小波分析法中的所有周期成分,且模型结果很好地验证了所提假设,可确定原序列存在5.6年的主要周期项。Aiming at the drawbacks of wavelet analysis method for cycle identification of annual runoff series,taking Wulong Station of Wujiang River for an example,this paper proposed a new method to identify the cycle of hydrological time series based on the Holt-Winters model inherent structure.By simulating the hydrological series,it can be seen that this method could determine the possibility of annual runoff series cycle appeared to achieve the purpose of cycle identification according to the fitting accuracy.In addition,the results are compared with that of wavelet analysis method.The results show that the Holt-Winters model well captured all the periodic components in the wavelet analysis method,and the model results was good enough to verify the proposed hypothesis,and it could determine that the original sequence has the main period of 5.6years.

关 键 词：Holt-Winters方法 周期识别 年径流序列 粒子群优化算法 小波分析法 

分 类 号：P338[天文地球—水文科学] P333.6[水利工程—水文学及水资源]