文档名：极值服从广义Pareto分布的扭转载荷外推方法研究
摘要：提出了一种极值样本服从广义Pareto分布(generalizedparetodistribution,GPD)函数的扭转载荷时域外推方法,基于极值样本的均值超出函数,首先确定一个区间范围,并以形状参数最小的均方误差为目标,通过自抽样法确定最佳阈值,再采用极大似然估计法对GPD函数的形状参数和尺度参数进行估计,获取扭转随机载荷谱的极值样本,服从GPD分布规律.以商用车驾驶室的稳定杆为研究对象,介绍了扭转载荷采集的方法,基于所提出的扭转载荷时域外推方法进行外推研究,并分别从穿级计数、功率谱密度、雨流图和潜在伪损伤量等方面对外推前后的载荷谱进行了对比分析.结果表明:所构建的载荷外推算法对商用车驾驶室稳定杆扭转载荷外推有较好适应性.
Abstract：Serviceloadspectrumsofthemechanicalstructurethroughoutitslifecyclearethefoundationofreliabilitydesign.Intheory,thesespectrumsaremeasurableviatheinstallationofthesensorsonthemechanicalstructure.However,inpractice,duetothelabourandR&Dcosts,theserviceloadspectrumsareonlymeasuredforashortperiodintheentirelifecycleofmechanicalstructure.Toobtaintheloadspectrumthroughoutthelifecycle,loadspectrumextrapolationtechniqueisusuallyemployedinpractice,whichinvolvesusingtheprobabilitystatisticallawofthemeasuredrandomloadspectrumtopredicttheremainingoperationalloadspectrumunderthesameworkingcondition.Currently,theloadspectrumextrapolationtechniqueisgenerallydividedintotwomaincategories:time-domainextrapolationmethodandrainflow-domainextrapolationmethod,accordingtotheformofextrapolation.Fortherainflow-domainextrapolationmethod,thefrom-toorrange-meanrainflowmatricesaregeneratedbycountingtheloadspectrumandareemployedtoestimatethetwo-dimensionaljointprobabilitydensityoftherainflowmatrixbasedonthekerneldensitytheory.Theparametersoftheprobabilitydensityfunctionhavebeenestimatedusingthemaximumlikelihoodestimationmethod.Inthistypeofextrapolationmethod,theaccuracyoftheextrapolatedloadspectrumisaffectedbytheselectionoftheprobabilitydensityfunctionandtheestimatedparametersoftheprobabilitydensityfunction.Forthetime-domainextrapolationmethod,statisticaltechniquesareadoptedtobuildthemeasuredloadspectrum'scharacteristicmodel.Theextremevaluetheory,forexample,isemployedtocharacterizetheprobabilitydistributionoftheexcesssamples.Thetime-domainextrapolationmethodextrapolatestheloadspectrumtoadesiredlengthwithagivenextrapolatedfactor.Moreover,itextrapolatestheloadspectrumdirectlywithoutanyconversions,suchasrainflowcountingandMarkovtime-domainreconstruction,whichreducestheerrorsgeneratedbytoomanylinksandmaximizestheretentionofloadinformation.Therefore,thetime-domainloadspectrumextrapolationmethodbasedontheextremevaluetheoryhasattractedgrowinginterestamongresearchersandbecomeapopularmethodinloadspectrumextrapolationresearch.Toaddressthemainweaknessesoftheexistingmehtod,atime-domainextrapolationmethodfortorsionalloadswithextremevaluesamplesfollowingthegeneralizedParetodistribution(GPD)functionisproposedinthispaper.Themainprocessoftheproposedextrapolationmethodissummarizedinthreesteps:turningpointextraction,extremevaluesampling,andextremesampledistributionfunctionfitting.Specifically,theextrapolationmethodfirstdeterminesanintervalrangebasedonthemeanvalueoftheextremevaluesamplesexceedingthefunction,andtakestheminimumMeansquarederroroftheshapeparametersasthegoal,determinestheoptimalthresholdvaluethroughtheselfsamplingmethod,andthenestimatestheshapeparametersandsizeparametersoftheGPDfunctionusingthemaximumlikelihoodestimationmethod.TheextremevaluesamplesobtainedfromthetorsionalrandomloadspectrumobeytheGPDdistributionlaw.Takingthestabilizerbarinthecabofcommercialvehicleastheresearchobject,thispaperintroducesthemethodoftorsionalloadacquisition,andthenconductsextrapolationresearchbasedontheproposedtorsionalloadtime-domainextrapolationmethod.Ourresultsindicatetheconstructedloadextrapolationmethodhasgoodadaptabilitytotheloadextrapolationofthetorsionalrotationofthecommercialvehiclecabstabilizerbar.
作者：郑国峰 陈柏先 隗寒冰 杨昊民 严璐瑶 Author：ZHENGGuofeng CHENBaixian WEIHanbing YANGHaomin YANLuyao
作者单位：重庆交通大学机电与车辆工程学院,重庆400074;中国汽车工程研究院股份有限公司,重庆401122重庆交通大学机电与车辆工程学院,重庆400074
刊名：重庆理工大学学报
Journal：JournalofChongqingInstituteofTechnology
年，卷(期)：2024, 38(3)
分类号：U463.8
关键词：扭转随机载荷 时域外推 广义Pareto分布 驾驶室稳定杆
Keywords：loadspectrums time-domainextrapolation generalizedParetodistribution stabilizerbar
机标分类号：STP391.4U463.81
在线出版日期：2024年3月25日
基金项目：中国博士后科学基金面上项目，国家自然科学基金，国家自然科学基金，重庆市教委科学技术研究计划项目，重庆市自然科学基金面上项目，重庆交通大学研究生科研创新项目极值服从广义Pareto分布的扭转载荷外推方法研究[
期刊论文] 重庆理工大学学报--2024, 38(3)郑国峰 陈柏先 隗寒冰 杨昊民 严璐瑶提出了一种极值样本服从广义Pareto分布(generalizedparetodistribution,GPD)函数的扭转载荷时域外推方法,基于极值样本的均值超出函数,首先确定一个区间范围,并以形状参数最小的均方误差为目标,通过自抽样法确定最佳...参考文献和引证文献
参考文献
引证文献
本文读者也读过
相似文献
相关博文
极值服从广义Pareto分布的扭转载荷外推方法研究 Research on the extrapolation method of torsional load with extreme values following generalized Pareto distribution
极值服从广义Pareto分布的扭转载荷外推方法研究.pdf
- 文件大小:
- 2.7 MB
- 下载次数:
- 60
-
高速下载
|
