DC　arc　faults　are　major　causes　of　electrical　fires　in　photovoltaic　(PV)　systems.　During　the　operation　and　maintenance　of　these　systems,　it　is　essential　not　only　to　identify　arc　faults　but　also　to　determine　their　exact　locations　accurately.　To　address　the　issue　of　DC　arc　fault　localization　in　PV　systems,　this　study　investigates　the　electromagnetic　radiation　(EMR)　characteristics　of　fault　arcs　and　proposes　a　method　for　DC　arc　fault　localization　using　the　redundant　antenna　array　and　the　ellipse　algorithm.　Firstly,　during　arc　combustion,　the　EMR　signals　collected　by　antennas　are　subjected　to　median　filtering　to　calculate　the　root　mean　square　(RMS),　which　serves　as　the　signal　strength.　An　artificial　neural　network　(ANN)　model　is　constructed,　which　uses　the　signal　strength　and　irradiance　to　predict　the　distance　between　the　fault　point　and　the　receiving　point.　Subsequently,　various　redundant　antenna　array　configurations　are　evaluated　to　assess　the　impact　of　different　antenna　quantities　and　layouts　on　localization　accuracy.　Once　the　optimal　layout　is　determined,　the　three　antennas　with　the　strongest　signal　are　selected.　Their　coordinates,　along　with　the　predicted　distances　to　the　fault　point,　are　input　into　the　ellipse　algorithm,　which　is　improved　by　trilateration,　to　obtain　the　locations　of　arc　faults.　Finally,　the　density-based　spatial　clustering　of　applications　with　noise　(DBSCAN)　method　is　used　to　fuse　multiple　measurement　results,　eliminate　interference,　and　confirm　the　final　fault　coordinates.　Experimental　results　demonstrate　that　the　proposed　location　method　exhibits　excellent　positioning　capability　and　adaptability,　with　an　average　positioning　error　of　0.365　m.