The　output　characteristics　of　single　screw　expander　has　a　direct　and　crucial　influence　on　the　performance　of　organic　Rankine　cycle　(ORC)　system.　In　this　paper,　a　machine　learning　prediction　model　driven　by　experimental　data　is　developed　and　applied　to　predict　the　power　output　of　single　screw　expander.　After　screening　different　structural　parameters　of　the　model,　genetic　algorithm　(GA)　is　used　to　optimize　the　initial　weights　and　thresholds　of　the　model,　so　as　to　further　improve　the　generalization　ability　of　the　model.　In　addition,　the　generalization　ability　of　the　model　is　compared　with　that　of　the　model　not　optimized　by　GA.　Furthermore,　the　influence　of　operating　parameters　on　the　power　output　of　single　screw　expander　is　analyzed　by　fitting　algorithm　in　three-dimensional　space.　The　optimization　boundary　value　needed　for　prediction　and　optimization　is　determined　by　fitting　algorithm　in　four-dimensional　space.　Finally,　a　prediction　and　optimization　model　is　created　by　coupling　the　machine　learning　prediction　model　with　GA,　and　the　maximum　power　output　and　corresponding　operating　parameters　of　the　single　screw　expander　under　full　operating　conditions　are　predicted　and　optimized.　The　results　show　that　with　the　application　of　machine　learning　and　GA,　the　maximum　power　output　of　single　screw　expander　can　be　predicted　and　optimized　precisely　under　full　operating　conditions.　So　as　to　directly　guide　the　selection　of　relevant　parameters　in　the　process　of　theoretical　analysis　and　experimental　research.