基于离散小波变换与时间序列的钱塘江潮位预测  

Forecasting Qiantang River Tide Level Based on Discrete Wavelet Transform and Time Series

作　　者：王培力[1] 王瑞荣[1] 高鹏[1] 朱广明[1] 余小庆[1] 孙映宏[2] 

机构地区：[1]杭州电子科技大学生命信息与仪器工程学院,浙江杭州310018 [2]杭州市水文水资源监测总站,浙江杭州310014

出　　处：《水电能源科学》2016年第6期6-8,13,共4页Water Resources and Power

基　　金：国家自然科学基金项目(61374005);浙江自然科学基金项目(LY14F030022)

摘　　要：针对钱塘江潮位呈现出的周期性、随机性和波动性,为实现对钱塘江潮位的有效预测,提出一种基于离散小波变换和时间序列的预测方法,即先利用离散小波变换将实测的钱塘江潮位序列进行分解与重构,将非平稳的序列转化为多层较平稳的序列;然后利用时间序列建模方法对分解后的各个序列分别建立时间序列模型,对各层进行动态预测;最后将各层预测值求和作为最终的预测结果。试验表明,所提方法预测的效果明显优于其他混合模型及单一模型,能够提供更加准确的潮位预测。The tide level of Qiantang River shows periodicity,randomness and volatility.In order to forecast the tide level of Qiantang River effectively,a forecasting method was proposed based on discrete wavelet transform and time series.Firstly,the discrete wavelet transform was used to make decomposition and reconstruction the original tide level time series,and several relatively stable time series was obtained.Then the time series models were adopted to establish each relatively stable tide level series and each sub-series was forecasted dynamically.Finally,multi-layer prediction values were added as the final prediction result.Experiment result shows that the prediction effect is much better than other hybrid models and single model,which can provide a more accurate tide level prediction for the Qiantang River.

关 键 词：钱塘江 潮位预测 时间序列 离散小波变换 混合模型 

分 类 号：TV11[水利工程—水文学及水资源]