【摘要】 准确的风电场风速预测对含大规模风电的电力系统安全稳定运行具有重要意义。针对时间序列法、卡尔曼滤波法、神经网络法等单一预测模型预测精度不高的问题,引入集成学习的分析方法,提出了一种基于Ada-boost算法改进的支持向量机(SVM)短期风速预测方法。该算法使用多个SVM模型通过加权累加得到最终输出,弥补了单一预测模型的缺陷。同时引入隶属度函数,通过赋予历史数据样本不同的权重来突出不同时间样本在预测模型中的差异性。以内蒙古风电场的实际采集数据为算例进行测试,结果表明模型预测精度显著提高,为实现更准确的在线短期风速预测提供了可能。更多还原

【Abstract】 Accurate wind speed prediction is important for safe and stable operation of power systems with large-scaleinterconnection of wind power. For a single time series prediction method,such as Kalman filtering,neural networkmodels,the prediction accuracy is not high,this paper proposes an ensemble learning predictive model based on Ada-boost algorithm and support vector machine(SVM)method for short-term wind forecasting. The algorithm utilizes multi-ple SVM models to obtain the final output by a weighted cumulative,to make up for the shortcomings of a single fore-cast model. While with the introduction of membership function,historical data have been given different weights tohighlight the difference in the prediction model. Treating wind farms in Mongolia region as a example,the results indi-cate that the proposed method significantly enhances the accuracy and to make a more accurate short-term wind forecast-ing online possible.更多还原