基于小波包变换和峰式马尔科夫链的风速短期预测  

Short-term wind speed forecast based on wavelet packet decomposition and peak-type Markov chain

作　　者：冯凯[1] 应展烽[1] 吴军基[1] 万萌[1] 

机构地区：[1]南京理工大学能源与动力工程学院,江苏南京210094

出　　处：《南京理工大学学报》2014年第5期639-643,657,共6页Journal of Nanjing University of Science and Technology

基　　金：江苏省普通高校研究生科研创新计划项目(KYLX_0370);连云港市产学研联合研究项目(CXY1216)

摘　　要：利用小波包的信号分析特性及峰式马尔科夫链的非稳态数据预测特性,提出一种新型风速短期预测方法。该方法对历史风速数据进行小波包分解,利用峰式马尔科夫链对小波包系数进行统计,分别得到小波包系数上升转移概率矩阵和下降转移概率矩阵,并由此得到下一时刻的小波包预测系数,将小波包系数进行重构可得到预测的风速。该文模型与其他模型的预测结果对比表明,该方法具有较高精度。A short-term wind speed forecast method is proposed here based on the signal analysis feature of the wavelet packet and the unsteady data forecast feature of the peak-type Markov Chain. This method decomposes the history wind speed data into the wavelet packet component and statistics these components with the peak-type Markov chain for establishing the rising probability matrices and the falling probability matrix. According to above matrices, the predicted wavelet packet coefficients at next moment are calculated and the wind speed is forecasted after reconstructing the predicted wavelet packet coefficients. The comparison between the proposed method and other forecast methods shows that the proposed method has the higher forecast accuracy.

关 键 词：小波包 峰式马尔科夫 风速短期预测 

分 类 号：TM614[电气工程—电力系统及自动化]