欢迎来到天天文库
浏览记录
ID:56029642
大小:405.54 KB
页数:6页
时间:2020-06-19
《基于相空间重构和遗传优化SVR的机械设备状态趋势预测.pdf》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、第34卷第3期噪声与振动控制Vol34NO.32014年6月NOISEANDVIBRAT10NC0NTR0LJun.2014文章编号:1006.1355(2014)03—0176-06基于相空间重构和遗传优化SVR的机械设备状态趋势预测王涛,李艾华,高运广,蔡艳平,王旭平(第二炮兵工程大学机电工程系,西安710025)摘要:针对机械设备振动信号序列的非线性、非平稳性特点,提出了一种基于相空间重构与遗传优化支持向量回归机的设备状态趋势预测方法。首先,采用相空间重构技术将一维振动信号时间序列转化成矩阵形式,自适应地选取特征,以相
2、点作为输入特征训练SVR预测器;然后应用自适应遗传算法对惩罚因子、不敏感系数以及高斯核宽度进行同步优化,自动获取最佳的建模参数;最后构建SVR预测模型,并将其应用于某机组振动信号预测。实验结果表明,无论是单步还是24步预测,本文所提遗传优化SVR模型的预测精度都要比标准SVR模型的预测精度高,说明该方法对机械设备的运行状态趋势具有较好的预测能力。关键词:振动与波;相空间重构;自适应遗传算法;支持向量回归;振动信号;趋势预测中图分类号:TH165~.3;TP206~.3文献标识码:ADOI编码:10.3969/j.issn.1
3、006.1335.2014.03.037ConditionTrendPredictionofMechanicalEquipmentsBasedonPhaseSpaceReconstructionandGeneticOptimizationSupportVectorRegressionWANGTao,LIAi-hua,GAOYun-guang,CAIYang-ping,WANGXu-ping(Dept.ofMechanicalandElectronicEngineering,TheSecondArtilleryEngineer
4、ingUniversity,Xi’an710025,China)AbgUaet:Aimingatthenonlinearandnon—stationarycharacteristicsofvibrationsignalsequenceofmechanicalequipments,aconditiontrendpredictionmethodformechanicalequipmentsisproposedbasedonphasespacereconstructionandgeneticoptimizationsupportv
5、ectorregression(SVR).Firstofall,aone—dimensionaltimeseriesofvibrationsignalsaretransformedintoamarxbyuseofphasespacereconstructiontechnique,anditsfeaturesareselectedadaptively.ThephasepointsareimportedtoSVRmodelasinputfeaturesandtheSVRpredictoristrained.Then,adapti
6、vegeneticalgorithmisappliedtooptimizethepenaltyfactorC,non-sensitivefactorandGaussiankemelwidthsynchronously.Thebestmodelparametersareobtainedautomatically.Finally,theSVRpredictionmodelisconstructedandisappliedtovibrationsignalpredictionofamachineunit.Theexperiment
7、alresultsshowthatwhetherforsingle-stepor24一stepprediction,thepredictionaccuracyoftheproposedgeneticoptimizationSVRishigherthanthatoftheconventionalSVR,indicatingthattheproposedmethodhasagoodabilityforpredictionoftheconditiontrendofthemechanicalequipments.Keyw硎s:vib
8、rationandwave;phasespacereconstruction;adaptivegeneticalgorithms;supportvectorregression(SVR);vibrationsignal;trendprediction收稿日期:2013—08.02由于表征机
此文档下载收益归作者所有