首页|基于改进FCM聚类的光伏电站出力场景特性研究

基于改进FCM聚类的光伏电站出力场景特性研究

扫码查看
为提高新型电力系统的运行安全可靠性,需要对光伏电站的出力特性进行量化评估,并从高不确定性的众多发电场景中掌握光伏电站的运行规律.为此,基于场景聚类和缩减原理对光伏出力进行分析.首先提出光伏出力特性的评价指标,包括波动性和出力效率2个一级指标和相应的二级指标;然后采用模糊C均值(fuzzy C-means,FCM)聚类算法,基于密度思想和距离最大、最小原则确定初始聚类中心,通过不断迭代得到不同场景下光伏出力的聚类结果.考虑到传统聚类算法存在局部收敛性,难以确定最优聚类数目,提出结合聚类有效性指标来确定聚类最优数目,进而采用基于概率距离的前推回代法对得到的聚类结果进行场景削减,最终得到光伏电站季节典型出力场景.最后基于广东省某光伏电站实际出力数据进行分析,验证所提指标和算法的有效性.
Study on Typical Output Scenario Characteristics of Photovoltaic Power Station Based on Improved FCM Clustering
In order to improve the operation safety and reliability of the new power system,it is necessary to quantitatively evaluate the output characteristics of the photovoltaic(PV)power station,and master the operation law of the PV power station from many power generation scenarios with high uncertainty.Therefore,this paper analyzes the PV output based on the theory of scene clustering and reduction.Firstly,it proposes the evaluation indexes of photovoltaic output characteristics,including two primary indexes,volatility and output efficiency,and corresponding secondary indexes.Then it uses the fuzzy C-means(FCM)clustering algorithm to determine the initial cluster center based on the density idea and the max-min principle of distance,and the clustering results of PV output in different scenarios are obtained through continuous iterations.Considering local convergence of traditional clustering algorithms,it is difficult to determine the optimal number of clusters,the paper proposes the clustering efficiency index to determine the optimal numbers of clusters,and then uses the the forward push back method based on probabilistic distance for clustering result reduction.Finally,the typical seasonal output scenarios of PV power stations are obtained.Through the analysis of the actual output data of a PV power station in Guangdong province,the effectiveness of the proposed index and algorithm is verified.

PV output scenarioclustering algorithmscene reductionFCM algorithmclustering index

苗璐、樊玮、肖红燕、刘宇、陈德扬、张勇军

展开 >

广东电网有限责任公司电力调度控制中心,广东 广州 510220

智慧能源工程技术研究中心,华南理工大学 电力学院,广东 广州 510640

光伏出力场景 聚类算法 场景缩减 FCM算法 聚类指标

中国南方电网有限责任公司科技项目广东省重点领域研发计划

036000KK52222013[GDKJXM20222142]2021B0101230001

2024

广东电力
广东电网公司电力科学研究院,广东省电机工程学会

广东电力

CSTPCD北大核心
影响因子:0.527
ISSN:1007-290X
年,卷(期):2024.37(3)
  • 27