摘要
本文提出了一种基于模糊聚类和支持向量机的光伏短期功率预测方法。通过气象信息建立模糊相似矩阵将光伏发电功率历史样本划分为若干类,然后通过分类识别获得与预测日最相似的一类历史日样本集,将其与预测日的气象因素作为预测模型的输入样本建立支持向量机光伏发电功率预测模型,并利用余一法对构建的支持向量机模型进行核参数和惩罚参数的优化。根据实际数据对所提模型进行验证,计算分析了预测误差,结果表明该方法具有较高的预测精度,对光伏发电预测具有一定的参考价值。
A short-term forecasting method for photovohaic (PV) power is proposed based on fuzzy clustering algorithm and support vector machine (SVM). Based on the meteorological information, fuzzy similarity matrix is established to divide the PV power generation history into several classes. The most similar history data to the forecasting day obtained by pattern recognition and the meteorological factors on the forecasting day are used as input to establish an SVM fore- casting model of PV power, which uses leave-one-out algorithm to optimize the Kernel parameter and penalty parame- ter. The model is validated by PV system data from a real project, and the forecasting error is calculated and analyzed. The results show that the proposed method has high accuracy, which can provide reference for the forecasting of PV power.
作者
于秋玲
许长清
李珊
刘洪
宋毅
刘晓鸥
YU Qiuling XU Changqing LI Shan LIU Hong SONG Yi LIU Xiaoou(Economic Research Institute, State Grid Henan Power Company, Zhengzhou 450052, China School of Electrical Engineering and Automation, Tianjin University, Tianjin 300072, China State Power Economic Research Institute, Beijing 102209, China)
出处
《电力系统及其自动化学报》
CSCD
北大核心
2016年第12期115-118,129,共5页
Proceedings of the CSU-EPSA
关键词
气象信息
模糊聚类
支持向量机
光伏功率
短期预测
meteorological information
fuzzy clustering
support vector machine (SVM)
photovohaic (PV) power
short-term forecasting