The　failure　of　the　flexible　DC　transmission　system　greatly　affects　the　stability　of　power　system.　The　existing　transmission　line　fault　detection　methods　have　the　problems　of　difficult　threshold　selection,　sensitivity　to　transition　resistance　changes　and　long　detection　time.　A　method　of　fault　type　detection　and　position　discrimination　based　on　wavelet　energy　ratio　using　PNN(Probabilistic　Neural　Network)　is　proposed.　The　frequency　characteristics　of　the　transient　voltage　are　obtained　by　fast　Fourier　analysis　of　the　measured　voltages　of　the　bus　and　line　under　different　fault　types,　and　then　DWT(Discrete　Wavelet　Transform)　is　used　to　obtain　the　wavelet　energy　characteristics　at　different　scales.　The　fault　type　and　fault　location　can　be　determined　accurately　according　to　the　output　results　of　PNN.　The　electromagnetic　transient　model　of　the　four-terminal　flexible　DC　transmission　network　is　built　under　PSCADEMTDC　environment.　The　simulative　results　show　that　the　proposed　method　can　detect　the　fault　type　and　fault　location　of　high　resistance　grounding　fault　accurately,　without　being　affected　by　the　transition　resistance.　©　2021　Electric　Power　Automation　Equipment　Editorial　Department.　All　right　reserved.