摘要
自然界中风固有的波动性直接影响风电功率的准确预测。因此,如何定量描述风电功率的波动性是解决该问题的关键。文中提出一种混合Logistic分布模型来定量描述风电功率的波动变化率,采用改进K均值聚类算法来确定模型参数。从不同采样间隔分布特性以及时间窗分布特性分析该模型性能,并将该分布模型与单一分布模型Normal分布、Logistic分布以及混合高斯分布等模型进行对比,通过利用吉林省某风电场的实测数据仿真实验,比较其评价指标,验证了该文提出模型的有效性。
Inherent fluctuation nature of wind adversely affects accurate prediction of wind power. Therefore, how to describe wind power volatility quantitatively is the key to solve this problem. A model based on mixed logistic distribution is proposed to describe variation of wind power. An improved K-means clustering algorithm is used to determine model parameters. By analyzing multi time scale distribution and time window characteristics of wind power, the distribution model is compared with single distribution model of normal distribution, logistic distribution and mixed Gaussian distribution. Effectiveness of the model is verified with simulation using measured data of a wind farm in Jilin Province.
出处
《电网技术》
EI
CSCD
北大核心
2017年第5期1376-1382,共7页
Power System Technology
基金
国家重点基础研究发展计划项目(973计划)(2013CB228201)
国家自然科学基金项目(51307017)
吉林省产业技术与专项开发项目(2014Y124)
吉林省科技发展计划(20140520129JH)~~
关键词
波动特性
混合Logistic分布
改进K均值聚类
多时间尺度
时间窗
fluctuation characteristics
mixed logistic distribution
improved K-means clustering
multi time scale distribution
time window
