基于自适应k均值和SVR的光伏出力预测
Photovoltaic Output Prediction Based on Adaptive k-means and SVR
孙艳玲 1朱晨光 1邵山 2田媛 2陈中杰 2谢东阳2
作者信息
- 1. 平高集团有限公司,河南 平顶山 467000
- 2. 平高综合能源服务有限公司,河南 平顶山 467000
- 折叠
摘要
为解决光伏功率预测不准确问题,提出了一种基于自适应k均值和支持向量回归的光伏出力预测方法.首先,分析了 k均值聚类及其改进方法,给出了支持向量回归(SVR)的基本原理和应用流程,介绍了 SVR中径向基函数凸优化模型.然后,结合自适应k均值和支持向量回归,依据光伏出力基本特点,分析了光伏出力预测流程及预测结果统计学评价指标.最后,以"云南昆明"光照数据为实际算例,确定了预测模型结构,并分别采用k-means and SVR、ARMA和ANN这3种方法进行预测,对比了不同聚类结果和不同算法时的预测统计指标,验证了所提方法的有效性,为光伏出力预测提供了一种方法.
Abstract
To solve the problem of inaccurate photovoltaic power prediction,a photovoltaic output pre-diction method based on adaptive k-means and support vector regression is proposed.Firstly,k-means clustering and its improvement methods are analyzed,and the basic principle and application process of support vector regression(SVR)are presented.The radial basis function convex optimization model is in-troduced.Then,combining adaptive k-means and support vector regression,based on the basic character-istics of photovoltaic output,the photovoltaic output prediction process and statistical evaluation indicators of prediction results are analyzed.Finally,taking the lighting data of"Kunming,Yunnan"as an actual cal-culation example,the prediction model structure is determined,and three algorithms,k-means and SVR,ARMA,and ANN,are used for prediction.The prediction statistical indicators under different clustering results and algorithms are compared to verify the effectiveness of the proposed method,providing a method for photovoltaic output prediction.
关键词
自适应k均值/光伏功率/出力预测/支持向量回归Key words
adaptive k-means/PV power/output prediction/support vector regression引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据