Acoustic　emission　(AE)　is　a　dynamic　non-destructive　detecting　technique,　and　could　be　a　promising　application　for　failure　detection　of　plasma　sprayed　coatings.　This　work　presents　a　thorough　study　for　detecting　and　classifying　different　failure　modes　under　Vickers　indentation　testing　of　8　wt%　Y2O3-stabilized　ZrO2　(8YSZ)　thermal　barrier　coatings　(TBCs)　by　using　acoustic　emission　technique.　Cluster　analysis　and　wavelet　packet　transform　(WPT)　were　used　to　characterize　the　acoustic　emission　signals　released　from　as-sprayed　TBCs　during　indentation　testing.　Back-propagation　(BP)　neural　network　was　used　to　study　the　effect　of　thermal　exposure　treatment　on　damage　evolution　of　8YSZ　coatings.　The　results　of　cluster　analysis　showed　that　there　were　three　types　of　AE　signals　associated　with　distinct　failure　modes　for　8YSZ　thermal　barrier　coatings　under　Vickers　indentation　testing.　Based　on　wavelet　packet　transform,　the　dominant　frequency　bands　of　these　three　types　of　signals　were　concentrated　at　levels　0-156.25　kHz,　156.25-234.375　kHz　and　312.5-390.625　kHz,　respectively.　Combined　with　scanning　electron　microscope　(SEM)　micrographs　of　residual　Vickers　impression,　it　can　be　found　that　8YSZ　coatings　have　undergone　three　processes:　elastic-plastic　deformation,　micro-cracks　propagation　and　debonding　at　the　splat　boundaries　during　the　indentation　testing.　The　BP　neural　network　identification　results　showed　that　after　high　temperature　thermal　exposure　treatment,　the　strain　tolerance　and　service　durability　of　8YSZ　thermal　barrier　coatings　decreased　while　the　more　fracture　signals　are　generated　during　the　indentation　process.　The　high　temperature　oxidation　environment　accelerated　the　degradation　of　the　plasma　sprayed　8YSZ　thermal　barrier　coatings.