In　order　to　pick　up　fault　features　effectively,　it　is　necessary　to　select　a　reasonable　sensor　position　in　vibration　measurement　of　a　defected　rolling　bearing,　it　is　one　of　the　main　study　goals　in　the　field　of　predictive　maintenance.　Due　to　the　complicated　structure　of　a　rolling　bearing,　the　numerical　simulation　method　was　utilized　for　its　dynamic　response　analysis.　Based　on　a　FEM　model　of　6307　rolling　bearing,　its　kinetic　feature　and　stress　distribution　of　each　part　were　obtained.　As　a　result,　the　impulse　functions　induced　by　different　local　defects　were　gained,　they　were　then　exerted　on　the　assembled　model　including　a　bearing　chock　as　excitations.　Considering　the　nonlinear　contact　characteristic　simultaneously,　the　vibration　response　on　each　node　of　the　whole　structure　was　obtained.　The　difference　between　the　response　waveform　of　the　node　on　the　chock　surface　and　the　bearing　vibration　signal　measured　in　tests　was　discussed　thoroughly　when　the　defect　was　located　on　different　bearing　parts.　Furthermore,　the　vibration　signal　of　the　rolling　bearing　with　the　seeded　fault　was　picked　up　to　validate　the　simulation　results.　Under　the　condition　that　the　FE　model　is　reliable,　the　transmission　process　of　the　vibration　response　in　each　part　was　discussed,　the　emphasis　was　focused　on　its　propagation　and　attenuation　characteristics　when　the　vibration　wave　passed　the　interface　of　the　outer　race　and　bearing　chock.　Based　on　the　above　theoretical　results,　the　optimal　sensor　position　and　measuring　direction　were　determined　through　comparing　the　response　feature　of　different　surface　nodes.