风速预测是风力发电领域的重要课题之一.风速波动剧烈,预测难度大,深入发掘风速数据波动性特征对于提高风速预测的准确性有积极意义.根据随机波动(SV)模型的峰度分析技术,研究风速时间序列的高峰度特征.基于电力系统领域对峰度的定义,理论推导并证明SV超峰度定理的衍生形式,建立适应风速预测的SV风速模型,模拟风速数据的整体峰度.在分析SV-t族模型的基础上,为选择适当的SV风速预测模型的条件分布类型提供了一种有效方案.实际风电场数据算例分析表明,该方法能有效建立高峰度特征的实际风速模型,对实际风速建模有一定的实用意义.%Wind speed forecasting is one of the most important issues in the domain of wind power generation. Wind speed changes irregularly and fluctuates violently, which makes it difficult to be forecasted. Comprehensive investigation on characteristics of the wind speed data is a positive key solution to improve the precision of wind speed forecasting. With the technique of stochastic volatility (SV) model kurtosis analysis, the leptokurtosis of wind speed was developed. Based on the specific definition of kurtosis in power systems, the extended SV excess kurtosis theorem was induced and proved. SV models were built adapting to the wind speed characteristics. In the case study, SV models with different conditional distributions were also presented and the overall kurtosis of wind speed was analyzed. In addition, a feasible scheme based on the SV-t type models for choosing the best conditional distribution in SV wind speed forecasting model was proposed. Results of the case study based on actual wind speed data demonstrate clearly that the proposed model is a valid method for wind speed modeling with the consideration of the wind characteristics of excess kurtosis and can provide a promising way for the practical wind speed modeling.
展开▼
