基于LARS特征选择的风电机组故障诊断的研究  

Research on fault diagnosis of wind turbines based on LARS feature selection

作　　者：孙群丽[1] 周瑛 刘长良[3] Sun Qunli;Zhou Ying;Liu Changliang(Science&Technology College,North China Electrie Power University,Baoding 071003,China;Sifang College,Shijjazhuang Tiedao University,Shijiahuang 051132,China;State Key Laboratory of Altemate Electrical Power System with Renewable Energy Sources,North China Eletric Power University,Beijing 102206,China)

机构地区：[1]华北电力大学科技学院,河北保定071003 [2]石家庄铁道大学四方学院,河北石家庄051132 [3]华北电力大学新能源电力系统国家重点实验室,北京102206

出　　处：《可再生能源》2020年第10期1349-1354,共6页Renewable Energy Resources

基　　金：中央高校基本科研业务费专项基金资助项目(9161717007);北京市自然科学基金项目(4182061)。

摘　　要：在风力发电机组运行维护过程中,对设备故障诊断的要求越来越高,随着近年来大数据的广泛应用,其对风力发电的影响也越来越大,许多研究人员基于大数据展开了相关工作。在利用风电机组SACDA数据进行故障诊断的过程中,所用的特征量不同,解决问题的效果会受到影响。为了提高风电机组故障诊断的精准性,须要对其所用到的故障特征进行选择。文章提出了用最小角回归(Least Angle Regression, LARS)方法来对特征向量进行选择,针对这些被选出的特征向量用HMM(Hidden Markov Model)建立故障模型。利用某风场的运行数据进行验证,结果表明,文章提出的基于HMM-LARS方法建立的模型对故障类型具有较好的识别效果。During the operation and maintenance of wind turbines,the requirements for equipment fault diagnosis are getting higher and higher.In addition to the widespread application of big data in recent years,its impact on wind power generation is also increasing,and many researchers have carried out related work based on big data.In the process of fault diagnosis using SACDA data of wind turbines,diferent feature quantities are used,and the effect of problem resolution will be afected.In order to improve the accuracy of wind turbine fault diagnosis,it is necessary to select the features used in fault diagnosis.This paper proposes using the LARS(Least Angle Regression)method to improve the effectiveness of feature selection,for these selected features,use HMM(Hidden Markov Model)to build a fault model.The simulation is performed on the operational data of a wind field,and the results show that the model proposed in this paper based on the HMM-LARS method has a better recognition efet on the fault types.

关 键 词：风电机组 故障诊断 隐马尔可夫模型 特征选择 最小角回归 

分 类 号：TK83[动力工程及工程热物理—流体机械及工程]