基于小波变换和LS-SVM的短期风速预测方法

SHORT TERM WIND SPEED PREDICTION BASED ON WAVELET TRANSFORM AND LS-SVM

针对风速序列的周期性和非平稳性,提出了基于小波变换和LS-SVM相结合的风电场风速预测模型,利用小波变换的多分辩分析法对风速序列进行分解,将风速序列投影到不同尺度上。结合LS-SVM的小样本学习能力强和计算简单等特点,将小波分解后的各风速子序列分别采用LS-SVM进行训练和预测,最后将各预测结果进行叠加得到最终的风速预测值。与LS-SVM风速预测方法进行比较,采用该文提出的方法可明显提高短期风速预测的精度,并具有较强的适应性,具有一定的工程应用前景。最后通过具体实例验证了该模型的有效性。

Aiming at the periodicity and non-stability of the wind speed change, a wind speed prediction model for wind farm based on wavelet and LS-SVM was proposed in this paper, in which wind speed sequence was decomposed by multi-resolution analysis method of wavelet transform and was reflected to different spaces. Combining with the characteristic of the strong learning ability and computation ability of LS-SVM＇ s for small sample, the sub-sequences after wavelet decomposing were trained and tested by LS-SVM and the final wind speed prediction values could be yielded by the linearity superposition for each prediction result. Compared with LS-SVM, the method proposed in this paper ean obviously improve the prediction accuracy of short term wind speed and has strong adaptability as well as quite bright prospect for engineering applications.