In　the　actual　production,　it　is　expected　to　find　several　sensitivity　geometric　errors　which　have　great　influence　on　machining　accuracy,　so　as　to　provide　references　for　the　manufacturing,　assembly,　and　other　links　of　the　machine　tool,　and　fundamentally　improve　the　working　performance.　To　study　this　problem,　a　novel　global　sensitivity　analysis　(GSA)　method　is　proposed.　Based　on　the　MBS　theory,　the　spatial　error　model　is　established　to　analyze　the　local　influence　of　geometric　error　on　machining　accuracy.　An　improved　second-order　partial　correlation　coefficient　based　on　the　Pearson　product　moment　is　proposed　to　analyze　the　correlation　between　geometric　errors.　In　addition,　the　error　parameters　will　change　greatly　in　the　stroke　of　the　moving　parts.　However,　the　rapid　fluctuation　of　geometric　error　value　will　have　a　dynamic　impact　on　the　machining　accuracy　of　machine　tools,　which　is　rarely　noticed　in　the　previous　sensitivity　analysis.　This　change　is　defined　as　the　fluctuation　of　error.　The　high　fitting　degree　function　of　error-displacement　is　obtained　by　using　the　high　order　Fourier　series　and　sine　series.　Then,　the　fluctuation　of　the　error　in　the　machining　stroke　is　analyzed　by　using　the　derivative　of　the　function　to　the　displacement.　Through　geometric　error　characteristics　(including　the　local　influence,　correlation,　and　fluctuation)　are　studied　comprehensively,　the　GSA　of　error　is　carried　out.　Finally,　taking　a　machining　center　as　an　example　and　combining　the　sensitivity　analysis　results,　the　improvement　measures　are　proposed　to　verify　the　correctness　of　the　method.　©　2021,　The　Author(s),　under　exclusive　licence　to　Springer-Verlag　London　Ltd.　part　of　Springer　Nature.