基于DTCWT与LSSVM的风电场短期风速预测

Short-term Wind Speed Forecasting of Wind Farms Based on DTCWT and LSSVM

由于风速具有很强的非线性特性,传统的预测方法难以对其准确预测。为提高预测精度,提出了将双树复小波与最小二乘支持向量机相结合的风速时间序列预测建模方法。首先,利用双树复小波对风速时间序列进行多尺度分解,将其分解为高频子带和低频子带;其次,利用最小二乘支持向量机对不同频率的子带建立相应的预测模型;最后,将各子带预测值进行等权求和得到预测结果。实验表明,基于双树复小波与最小二乘支持向量机的混合预测模型具有较高的预测精度,其平均绝对误差为3. 79%。

Because of the strong non-linear characteristics of wind speed, it is difficult to construct the model for accurate forecast with traditional methods. In order to improve the forecast precision, a forecasting method based on double tree complex wavelet transform(DTCWT) and least square support vector machine (LSSVM) is proposed in this paper. Firstly, DTCWT is used to depose wind speed time series into several high-frequency and low-frequency subbands. Secondly, LSSVM is used to establish the corresponding prediction model for different frequency subbands. Finally, these forecasting results of each subband are combined to obtain forecasting results. Experiments show that the hybrid forecasting model based on DTCWT and LSSVM has higher prediction accuracy and it is suitable for short-term wind speed forecasting of wind farms. And the value of MAPE can reach to 3.79%.