A　photovoltaic　(PV)　plant　is　normally　built　in　a　fixed　series-parallel　configuration　and　its　power-voltage　characteristics　often　get　complex　with　multiple　peaks　under　partial　shading　scenarios.　Therefore,　identification　of　partial　shading　is　important　for　monitoring　and　invoking　maximum　power　pint　estimation.　This　paper　proposes　a　back-propagation　neural　network　(BPNN)　based　partial　shading　identification　method　which　locates　shaded　modules　by　using　measured　voltage　data.　Optimal　sensor　placement　schemes　are　introduced　to　decrease　the　number　of　utilized　voltage　sensors,　and　meanwhile　still　keep　a　high　identification　performance.　Experiments　are　conducted　to　evaluate　the　accuracy　and　effectiveness　of　the　proposed　identification　method.　©　2018　IEEE.