资源描述:
《基于小波算法与时间序列法德风速预测改进方法》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、第41卷第1期中南大学学报(自然科学版)Vol.41No.12010年2月JournalofCentralSouthUniversity(ScienceandTechnology)Feb.2010基于小波分析法与滚动式时间序列法的风电场风速短期预测优化算法刘辉,田红旗,李燕飞(中南大学交通运输工程学院,轨道交通安全教育部重点实验室,湖南长沙,410075)摘要:为实现风电场风速的超前多步高精度预测,提出一种基于小波分析法与滚动式时间序列法混合建模的优化算法。该优化算法引入小波分析法对风电场实测非平稳风速序列进行分解重构计算,将非平稳性原始风速序列转化为多层较平稳分解风
2、速序列,利用对传统时间序列分析法改进后的滚动式时间序列法对各分解层风速序列建立非平稳时序预测模型,并通过模型方程实现超前多步滚动式预测计算。仿真结果表明:该优化算法实现了风速的高精度短期多步预测,将传统时间序列分析法对应超前1步、3步、5步的预测精度分别提高了54.22%,26.44%和19.38%,其预测的平均相对误差分别为1.14%,3.06%和4.41%;优化算法具有较强的细分与自学习能力。关键词:风速预测;滚动式时间序列法;小波分析法;时间序列分析法;优化算法中图分类号:TK81文献标志码:A文章编号:1672−7207(2010)01−0370−06Shor
3、t-termforecastingoptimizationalgorithmforwindspeedfromwindfarmsbasedonwaveletanalysismethodandrollingtimeseriesmethodLIUHui,TIANHong-qi,LIYan-fei(KeyLaboratoryofTrafficSafetyonTrack,MinistryofEducation,SchoolofTraffic&TransportationEngineering,CentralSouthUniversity,Changsha410075,China
4、)Abstract:Inordertoachievehigh-precisionmulti-stepaheadpredictionforreal-timewindspeeddatathatsampledfromwindfarm,basedonwaveletanalysismethodandrollingtimeseriesmethod,aforecastimprovedalgorithmwasproposed.Waveletanalysismethodwasusedtomakedecompositionandreconstructioncalculationsforo
5、riginalwindspeedseries,andmulti-layermoresteadywindspeedserieswasobtained.Thenrollingtimeseriesmethodthatwasmodifiedfromtraditionaltimeseriesmethodwasusedtobuildunsteadypredictionmodelsforeachlayer,andcorrespondingequationswereusedtorealizemulti-steprollingforecastcalculation.Simulation
6、resultsshowthattheoptimizationalgorithmattainshigh-precisionmulti-stepaheadforecastresults,improvesforecastprecisionofone-step,three-step,five-stepaheadforecasttraditionaltimeseriesmethodby40.48%,29.22%,45.73%,respectively,andthemeanrelativeerrorisonly1.72%,3.61%,7.12%,respectively.Theo
7、ptimizationalgorithmhasrespectivelyexcellentsubdivisionandself-learningability.Keywords:windspeedforecast;rollingtimeseriesmethod;waveletanalysismethod;timeseriesmethod;optimizationalgorithm收稿日期:2008−12−05;修回日期:2009−04−02基金项目:国家“十一五”科技支撑计划重点项目(2006BAC07B03);国家留学基金资助项目(200963706