Flexiblememristor　is　one　of　the　most　promising　wearable　devices　for　abundant　data　storage　and　processing.　In　this　work,　interface　engineering　by　inserting　the　Al2O3　barrier　layer　is　carried　out　to　construct　Pt/TiOx/Al2O3/Pt/indium　tin　oxide　(ITO)　flexible　artificial　synapse　device.　The　memristor　performance　can　be　maintained　even　under　1000　times　of　bending　without　degradation,　demonstrating　its　excellent　mechanical　property.　With　the　Al2O3　diffusion　barrier　layer,　the　oxygen　vacancies　(V-o)　movement　is　slowed　down　for　filaments　formation　and　rupture,　thus　it　boosts　up　the　synaptic　plasticity,　including　long-term　potentiation/depression,　paired-pulse　facilitation　(PPF),　and　spike-timing-dependent　plasticity　(STDP).　Moreover,　on　the　basis　of　the　enhanced　symmetry　and　linearity　of　conductance　for　Pt/TiOx/Al2O3/Pt/indium　tin　oxide　(ITO)　memristor,　the　neural　network　simulation　for　supervised　learning　presents　an　online　learning　pattern　recognition,　the　accuracy　can　achieve　to　91.15%.　Overall,　the　Pt/TiOx/Al2O3/Pt/ITO　memristor　with　excellent　flexibility　is　a　promising　emulator　for　biological　synapses,　which　could　be　beneficial　to　future　flexible　memristor-based　neuromorphic　computing.