基于小波分析和相空重构相融合的风速预测

Wind Speed Forecasting Model Based on Wavelet Analysis and Support Vector Machine

研究风速准确预测问题,针对风速具有随机性、间歇性以及混沌性,且含有大量噪声,造成预测精度误差。为提高风速预测精度,提出一种小波分析和相空间重构相融合的风速预测方法。首先采用小波技术消除风速历史数据,然后用混沌理论对数据重构,最后数据输入到支持向量机训练,并采用遗传算法优化支持向量机参数,建立最优风速最优预测模型。结果表明,相对传统预测方法,小波分析和相空间重构预测方法更能准确刻画风速的复杂变化特点,提高了风速的预测精度,也为其它非线性预测问题提供了一种新的研究思路。

The data of wind speed has chaotic, and contain a lot of noise, in order to improve the forecasting ac- curacy of wind speed, this paper proposes wind speed forecasting method based on wavelet analysis and support vector machine. Firstly, the data of wind speed are de-noised by wavelet analysis, then the data are reconstructed by chaot- ic theory, and finally, the data are input to support vector machine to learn, and the parameters of support vector ma- chine are optimized by to establish the optimal forecasting model of wind speed. The results show that, compared with the traditional methods, the proposed method can more accurately describe the change trends of the wind speed, and has improved the forecasting accuracy of wind speed, and it provides a new research idea for the nonlinear fore- casting problem.