摘要
准确的区域光伏功率预测作为解决光伏并网消纳和多能互补问题的技术之一受到越来越多的关注,提出一种基于典型代表电站和改进支持向量机(SVM)的区域光伏功率短期预测方法。通过K-means聚类将同一地区光伏电站划分到不同汇聚区,使用历史数据和3种数学相关系数计算得到各汇聚区典型代表电站,并通过4类光伏功率指标分析各典型代表电站与汇聚区的一致性,基于此,以改进SVM代替传统的滚动预报形成区域功率预测模型。实际算例分析表明,所提方法可提升区域光伏功率短期预测精度。
Accurate regional photovoltaic power forecasting attracts more and more attention since it is one of the techniques for solving problems of photovoltaic grid-connection consumption and multi-energy complementary.A short-term forecasting method for regional photovoltaic power based on typical representative power stations and improved SVM(Support Vector Machine)is proposed.The photovoltaic power stations in the same region are divided into different convergence areas by K-means clustering,the typical representative station in each convergence area is calculated by using historical data and three mathematical correlation coefficients,and the consistency of each typical representative station with the convergence area is analyzed through four photovoltaic power indices.On this basis,a regional power forecasting model is formed by substituting the traditional rolling forecasting with the improved SVM.The actual example analysis shows that the proposed method can improve the short-term forecasting accuracy of regional photovoltaic power.
作者
张扬科
李刚
李秀峰
ZHANG Yangke;LI Gang;LI Xiufeng(Institute of Hydropower System&Hydroinformatics,Dalian University of Technology,Dalian 116024,China;Yunnan Electric Power Dispatching and Control Center,Kunming 650000,China)
出处
《电力自动化设备》
EI
CSCD
北大核心
2021年第11期205-210,共6页
Electric Power Automation Equipment
基金
国家自然科学基金资助项目(51879030)。
关键词
K-MEANS聚类
典型代表电站
短期预测
新能源出力
SVM
K-means clustering
typical representative power stations
short-term forecasting
renewable energy output
SVM