摘要
针对风电场故障预警研究,首先基于NJW(Ng-Jordan-Weiss)谱聚类和动态时间规整算法结合,对风场内风机进行机群划分,然后对某一机群中任意一台风机使用Elman神经网络建模来代替一个机群内所有的单机模型,最后将模型预测误差与滑动窗口技术结合,确定机组正常运行阈值,判断同一机群内其他机组所处运行状态。以河北风场SCADA数据为实例进行故障预警,结果显示该风场可以划分为3个机群,对机群1其中一台风机进行建模,另一台风机进行故障预警验证,结果显示该模型可以提前11 h预警故障。采用该方法可以有效简化风电场故障预警的复杂性,帮助运维人员提高工作效率,对智慧风场的推进具有积极意义。
For wind farm fault early warning research,firstly,based on the combination of NJW spectral clustering and dynamic time warping algorithm,the wind turbines in the wind farm are divided into clusters.Then,Elman neural network modeling is used to replace all the single-unit models for any wind turbine in a certain cluster.Finally,the model prediction error is combined with the sliding window technology to determine the normal operating threshold of the unit and judge the operating state of other units in the same cluster.The SCADA data of Hebei Wind Farm is used as an example for fault early warning,and the results show that the wind farm can be divided into three clusters.One of the wind turbines in cluster 1 is modeled,and the other wind turbine is tested for fault early warning.The test results show that the model could warn the fault 11 hours in advance.This method can effectively simplify the complexity of wind farm fault early warning,help operation and maintenance personnel improve work efficiency,and have positive significance for the promotion of smart wind farm.
作者
施萌
马永光
SHI Meng;MA Yongguang(School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,China)
出处
《电力科学与工程》
2020年第8期25-29,共5页
Electric Power Science and Engineering
作者简介
施萌(1996—),女,硕士研究生,研究方向为风电机组故障预警;马永光(1964—),男,教授,主要研究方向为仿真与故障诊断中的应用。
