基于概率测度变换的风速时间序列建模方法  

A Wind Speed Time Series Modelling Method Based on Probability Measure Transformation

作　　者：张宏宇[1] 印永华[1] 申洪[1] 张明[2] 王皓怀[1] 

机构地区：[1]中国电力科学研究院,北京市100192 [2]东北电力大学电气工程学院,吉林省吉林市132012

出　　处：《电力系统自动化》2013年第2期7-10,17,共5页Automation of Electric Power Systems

基　　金：国家高技术研究发展计划(863计划)资助项目(2011AA05A103)~~

摘　　要：结合中国东北某实际风电场多年测风数据,分析风电场风速的统计特性,得出多年月内同一时刻风速同样服从Weibull分布。通过概率测度变换衔接实际风速序列与回归分析模型时间序列,基于自回归滑动平均(ARMA)模型,建立单一风电场风速时间序列模型;基于向量自回归(VAR)模型,给出了不同风电场间风速相关性考虑方法,分别给出了模拟风速时间序列的模型及参数。通过对比得知,实际风电场风速数据与模拟得到的风速时间序列具有较好的一致性,基于概率测度变换构建模拟风速时间序列是可行、有效的。The real wind speed datum of wind farms in Northeast China are analyzed.The same hourly wind speed of the month in the years obeys a Weibull distribution.The actual wind speed series and the time-series of regression analysis model are connected by probability measure transformation.The model of a single wind farm wind speed is built by autoregressive moving average（ARMA）model,and multi-wind speed model built by vector auto regression（VAR）model.The detailed model parameters are shown.The actual wind speed data and simulated wind speed time series have a better consistence.The simulated wind speed time series based on the probability measure transformation are feasible and effective.

关 键 词：风速时间序列 测度变换 自回归滑动平均(ARMA) 向量自回归(VAR) 相关性 

分 类 号：TM614[电气工程—电力系统及自动化]