基于小波分解的统计模型对小时风速的预测  

Hourly wind speed prediction based on wavelet decomposition and statistical model

作　　者：赵慧[1] 党张利 赵志远[1] 王聚杰[1] 张文煜[1] 

机构地区：[1]兰州大学大气科学学院甘肃省干旱气候变化与减灾重点实验室,兰州730000

出　　处：《兰州大学学报（自然科学版）》2014年第4期513-516,共4页Journal of Lanzhou University(Natural Sciences)

基　　金：国家自然科学基金项目(41225018);国家重大科学研究计划项目(2011CB706900)

摘　　要：为提高小时风速的预测精度,提出了基于小波分解和AR模型的混合模型(WD-AR).模型应用小波分解技术将风速序列进行多层分解,再利用AR模型分别对各分解层的风速序列进行预测,最后将预测结果叠加得到预测值.采用河西地区风速观测数据对模型进行分析验证,结果表明:WD-AR模型预测精度指标R,RMSE和MAP E值分别是0.89,0.36和27%,与AR模型相比有了较大的改善,提高了小时风速的预测精度,说明WD-AR模型具有更好的预测能力.In order to improve the accuracy of hourly wind speed prediction, a hybrid model based on wavelet decomposition combined with AR model（WD-AR） was proposed, in which the wind speed sequence was first decomposed in a multilayered manner by wavelet decomposition technique. Then the auto regressive theory was used to build a prediction model for each layer. Finally, the wind speed prediction values could be yielded by the linearity superposition for each prediction result. Real data on wind speed from Hexi were used to verify the model. The results show that the WD-AR was greatly improved when compared with AR model, the prediction accuracy of the model parameters, i.e. R, RMSE and MAP E were 0.89, 0.36 and 27%. The conclusion here is that the WD-AR model is able to improve the accuracy of hourly wind speed prediction and has better predictive ability.

关 键 词：小波分解 AR模型 WD-AR混合模型 风速预测 

分 类 号：P413.2[天文地球—大气科学及气象学]