风电场概率分布模型建模及误差分析  

Modeling and Error Analysis of Wind Farm Probability Distribution Model

作　　者：梁海峰[1] 曹大卫[1] 刘博[1] 刘子兴 郑灿[1] 李鹏[1] 

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003 [2]国网安新县供电公司,河北安新071600

出　　处：《华北电力大学学报（自然科学版）》2017年第3期8-14,共7页Journal of North China Electric Power University：Natural Science Edition

基　　金：国家高技术研究发展计划(863计划)资助项目(2015AA050104)

摘　　要：提出一种风电场出力概率分布模型建立与误差分析的方法。基于典型日的历史风速和对应的风机无限制出力情况下的功率数据,拟合得到风机等效功率特性曲线,进而得到整个风电场无限制出力情况下的实际可发出力。利用高斯混合模型对风电场出力概率分布进行建模,采用序列运算理论,制定最佳离散化步长的选取规则,提高了概率密度函数拟合的精确性。提出基于误差范围的概率密度函数可靠性指标,利用随机抽样方法,对概率密度函数进行误差分析。通过与核密度估计方法进行对比,证明了高斯混合模型的优越性。所建模型能够较好地描述风电场出力的概率特性,为运行人员在风电调度运行方面提供有价值参考依据。In this paper, a method for establishing a model of wind farm output probability distribution and error analysis is proposed. Based on recorded wind speed data and the corresponding wind turbine output data without limitation in typical days, the equivalent power characteristic curve of wind turbine is obtained. And then the actual available output of the whole wind farm without limitation is got. The output probability distribution model of wind farm is established by using Gaussian mixture model. Through the sequence operation theory and the selection of optimal discretization step size, the accuracy of probability density function fitting is improved. The reliability index of probability density function basing on error range is proposed and the error of the probability density function is analyzed by using the random sam- piing method. The superiority of Gaussian mixture model is proved by comparing with kernel density estimation method. The model established in this paper well described the probabilistic characteristics of wind farm output, and provide valuable reference for operation personnel in wind power dispatching and operation.

关 键 词：风电功率 概率分布 高斯混合模型 序列运算理论 随机抽样 

分 类 号：TM614[电气工程—电力系统及自动化]