物理-监测信息融合的振动传递路径分析方法

打开文本图片集

关键词：传递路径分析；物理信息神经网络；悬臂梁；多源激励 中图分类号： TH123+.1 ; TH17 文献标志码：A DOI：10.16385/j.cnki.issn.1004-4523.202504038

Abstract：Trditioaltasferpathaalysis）etdsrequirefiedsesorplacementtoapturevbratioaracteristics，wilephical spaceconstraintsorarshnviromentalconditiosinpracticalengineringoftenpreventmeasurementsatiticalodes，significantly compromisingaalytcalaacyodessthisliitation，thisstudyproposesasics-monitoingintegatedbrationtraerpath analysismethod.Aphysics-iforedneuralnetwork（PINN）modelisonstructedbyembeddingelastodynamicequationsintoneural networks，andavbrationresponsepredictionfrmeorkisdeveloedundertheconstratsofgoveingdiferentialequatios，cobiing partialsurfacaeleratimeasueents.Aumcalslatiofthdimesialantilereasstisouctedtevatete proposedethod'spathidentificatioperformaneithonventioalTA，ndexperimentalesutssowtatteproosedmetodbtains peak frequency relative error of 0.15% andabsolute amplitude error of 6.19dB，outperforming traditional TPA and serving as a viable substitute.Near-feldeergystreamlientangleentarecitationsourcescanbeevealedbypowerfowfeldnalysis，dmantrasfer pathscanbeientifdandortexiudeedipatatfroundararactedissdyotoyfectielyo measurementlimitatiosatciticalnodes，butasoprovidesanovelmetodologyforanalyzingmulti-soucevibrationenergytraii mechanisms in complex mechanical systems.

Keywords：transfer pathanalysis;physical information neural network;cantilever beam;multi-source incentives

随着装备系统向轻量化、高速化与集成化方向不断发展，多物理场耦合激励下，振动传递路径日益复杂，传统基于简化假设的分析方法已难以满足高保真建模与精准调控的需求。（剩余13725字）