基于优化FCM聚类的RELM风速预测  

Wind Speed Forecasting of Regularized ELM Based on Optimized FCM Clustering

作　　者：潘超[1] 秦本双 何瑶 袁翀 沈清野 

机构地区：[1]东北电力大学电气工程学院,吉林省吉林市132012 [2]国网淳安供电公司,浙江省淳安市311700 [3]国网舟山供电公司,浙江省舟山市316000

出　　处：《电网技术》2018年第3期842-848,共7页Power System Technology

基　　金：国家863高技术基金项目(SS2014AA052502);国家自然科学基金项目(51507027)~~

摘　　要：准确的风速预测对大规模风电并网具有重要意义。提出一种基于互信息属性约简优化聚类的正则化极限学习机短期风速预测方法。首先考虑不同属性特征对风速的不同影响,计算风速特征属性序列与风速序列的互信息,并运用最大相关最小冗余算法进行特征选择,然后采用优化的模糊C均值聚类方法对风速样本进行聚类,再对极限学习机进行优化,进而构建风速组合预测模型。最后结合风电场实测数据进行风速预测实验,结果表明该方法具有较高的预测精度。Accurate wind speed forecasting is of great significance for large-scale wind power integration.In this paper,a new method of short-term wind speed forecasting is put forward based on regularized extreme learning machine（ELM） of optimal clustering and mutual information attribute reduction.Firstly,considering different effects of different attributes on wind speed,mutual information between wind speed characteristic sequence and wind speed sequence are calculated.The attribute features are selected with maximumcorrelation minimum-redundancy algorithm.Then,wind samples are clustered with optimized fuzzy C-means（FCM） clustering method.The ELM is optimized and a combined forecasting model of wind speed is constructed.Finally,wind speed prediction experiment is carried out combined with measured data of wind farm.Results show that the method has high prediction accuracy.

关 键 词：风速预测 最大相关最小冗余 模糊C均值聚类 正则化 极限学习机 

分 类 号：TM614[电气工程—电力系统及自动化]