文章编号：1000-8055（2023）04-0931-08doi：10.13224/j.cnki.jasp.20210516基于LASSO变量选择的航空发动机相似性剩余寿命预测于倩影1，李娟1，戴洪德2，辛富禄1（1.鲁东大学数学与统计科学学院，山东烟台264025；2.海军航空大学航空基础学院，山东烟台264001）摘要：由于航空发动机监测变量众多，传统方法直接选取性能退化趋势明显的变量进行寿命预测，所以提出一种基于LASSO（leastabsoluteshrinkageandselectionoperator）的变量选取方法，结合相似性寿命预测方法有效提高了预测精度。基于K-means聚类区分不同工况，对航空发动机多个监测变量根据聚类结果进行变量转换。基于LASSO方法选取最优传感器变量。基于相似性方法进行航空发动机剩余寿命预测。将基于LASSO的变量选取方法与传统的根据退化趋势大小进行选择的方法进行剩余使用寿命预测的结果进行了对比研究。结果表明：基于LASSO选取变量的相似性寿命预测误差的标准差在3种运行周期下分别减少了约1.84、3.46、4.23。关键词：预测与健康管理；K-means聚类；LASSO方法；相似性；剩余寿命中图分类号：V263.6文献标志码：ALASSObasedvariableselectionforsimilarityremainingusefullifepredictionofaero-engineYUQianying1，LIJuan1，DAIHongde2，XINFulu1（1.SchoolofMathematicsandStatistics，LudongUniversity，YantaiShandong264025，China；2.SchoolofBasicSciencesforAviation，NavalAviationUniversity，YantaiShandong264001，China）Abstract:Duetothelargenumberofaero-enginemonitoringvariables,thevariableswithobviousperformancedegradationtrendweredirectlyselectedbytraditionalmethodforthelifeprediction,soavariableselectionmethodbasedonLASSO(leastabsoluteshrinkageandselectionoperator)wasproposed,whichcombinedwiththesimilaritylifepredictionmethodtoeffectivelyimprovethepredictionaccuracy.BasedonK-meansclustering,differentworkingconditionsweredistinguished,andmultiplemonitoringvariablesofaero-engineweretransformedaccordingtotheclusteringresults.TheoptimalsensorvariableswereselectedbasedontheLASSOmethod.Theremainingusefullifeofaero-enginewaspredictedbasedonsimilaritymethod.TheresultsofremainingusefullifepredictionbasedonthevariableselectionmethodbyLASSOandthetraditionalselectionmethodbythedegradationtrendwerecompare...
