Distribution　system　structures　are　intricate　and　variable,　with　single-phase　ground　(SPG)　fault　occurring　frequently.　Existing　section　location　methods　for　SPG　faults　are　often　influenced　by　manually　set　thresholds,　resulting　in　low　accuracy.　This　paper　analyzes　the　relationship　between　the　zero-sequence　voltage　and　zero-sequence　current　synthesized　from　three-phase　voltage　and　current　under　the　condition　of　no　zero-sequence　transformer　installed　on　the　line,　and　proposes　a　SPG　fault　section　threshold-free　location　method　based　on　Swin-Transformer.　Firstly,　Fourier　transform-based　data　processing　methods　correct　phase　errors　caused　by　current　transformers　(CT)　and　eliminate　the　DC　bias　in　synthesized　zero-sequence　currents.　Secondly,　the　synthesized　zero-sequence　current　waveform　and　the　derivative　of　zero-sequence　voltage　waveform　are　overlaid　on　the　same　image,　with　the　enclosed　areas　shaded　to　highlight　the　relationship　between　the　two　electrical　quantities.　Finally,　the　Transformer　model　is　used　to　classify　the　differentiated　images　between　the　SPG　fault　section　and　sound　section　to　achieve　threshold-free　fault　location.　The　proposed　method　has　been　tested　and　validated　on　an　industrial　prototype　in　a　10　kV　full-scale　test　system　and　deployed　in　actual　distribution　network.　Experimental　results　demonstrate　the　method　can　accurately　locate　fault　sections　even　under　conditions　of　noise　interference　and　inaccurate　trigger　timing.　©　1963-2012　IEEE.