摘要
提出了一种基于连续时间段聚类的支持向量机风电功率预测方法。通过2次聚类把全年分为若干个类型的连续时间段,并对同类型时间段使用支持向量机建模,建立后的模型用于其他年份对应时间段的预测。与神经网络相比,支持向量机建模方法避免了局部最优。利用国内某风电场数据进行对比实验,证明了所述方法的有效性。
A wind power prediction method based on sequential time clustering support vector machine (SVM) is proposed. Each year is divided into several continuous time series by clustering twice, with one catching the daily similarity to build model by SVM and the proposed model is used to predict statistics of time series corresponding to other years. The SVM model can avoid converging into a local optimal zone compared to neural networks method. Experiments on a wind farm show the effectiveness of the proposed method.
出处
《电力系统自动化》
EI
CSCD
北大核心
2012年第14期131-135,149,共6页
Automation of Electric Power Systems
关键词
风力发电
功率预测
时间段聚类
支持向量机
wind power generation
power prediction
sequential time clustering
support vector machine (SVM)