基于灰色聚类的电能质量综合评估方法 A power quality comprehensive evaluation method based on grey clustering.pdfVIP

第40 卷 第15 期 电力系统保护与控制 Vol.40 No.15 2012 年8 月1 日 Power System Protection and Control Aug. 1, 2012 支持向量机向量维数对短期风电功率预测精度的影响 王 宁，叶 林，陈 盛，朱庆伟 （中国农业大学信息与电气工程学院，北京 100083） 摘要：为了提高短期风电功率的预测精度，采用支持向量机回归方法，以 10 min 采样间隔的实测风速和温度序列为输入数 据，针对连续 19 天的历史数据按每隔 5 h 进行一次预测，每次采用一天 144 组风速、温度数据进行训练。实验分别选取了 10 min、30 min 和 1 h 的预测时间尺度，得到不同向量维数下预测结果的均方误差、均方百分比误差、平均绝对误差、平 均绝对百分比误差以及相关系数。结果表明，对于特定的预测对象，支持向量机向量维数具有一个使预测精度最高、误差最 小的最优值；对于不同的预测时间尺度，向量维数的最优值一般不同。 关键词：风力发电；短期功率预测；支持向量机；向量维数；预测精度 Influence of the vector- dimension of SVM on short-term wind power prediction accuracy WANG Ning ，YE Lin ，CHEN Sheng，ZHU Qing-wei (College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China) Abstract: In order to improve prediction accuracy of short-term wind power forecasting, t his paper presents a novel approach of Support Vector Regression (SVR) in short-term w ind power forecasting by u sing measured temperatures and wind speed data of 10 minutes sampling interval. For historical data of 19 days, prediction is made once every five hours, and each t ime 144 groups of temperatures and w ind s peed data in a d ay ar e u sed for tr aining. Experiments are carried out ba sed o n 10 min, 3 0 min and one hour time s cales respectively. M SE ，MSPE ，MAE ，MAPE an d c orrelation co efficient ar e al so cal culated u nder varying vector-dimension. It can be seen that there is an optimal vector-dimension which has the highest accuracy and minimum error for specific prediction. Results show that for different time scales, the optimal value of the vector dimension is generally different. This work is supp