Recently,　the　photovoltaic-driven　variable　frequency　air　source　heat　pump　(PV-VFASHP)　space　heating　system　has　been　widely　used　in　low-carbon　buildings.　However,　owing　to　the　lack　of　performance　prediction　model　for　VFASHPs,　the　self-consumption　rate　of　PV　power　generation　is　far　lower　than　expected.　To　solve　this　problem,　a　simulation　model　and　a　test　rig　for　VFASHP　were　built,　and　its　heating　performance　under　different　conditions　was　investigated.　A　performance　prediction　model　for　heating　capacity　and　COP　was　developed　considering　the　joint　effect　of　outdoor　temperature,　supply　water　temperature,　and　compressor　speed.　Then,　this　model　was　applied　to　a　PV-VFASHP　heating　system　in　a　low-carbon　building　in　Beijing,　and　the　application　effect　on　sunny,　cloudy,　and　rainy　days　was　analyzed.　Results　show　that　the　performance　prediction　model　can　predict　the　heating　capacity　and　COP　accurately.　Experimental　results　of　13　VFASHPs　from　8　manufacturers　show　that　the　relative　errors　are　all　within　+10　%.　When　regulating　the　PV-VFASHP　system　based　on　this　performance　prediction　model,　the　self-consumption　rate　of　PV　power　reaches　100　%　on　all　the　three　days,　contributing　77.26　%,　73.77　%,　and　67.84　%　of　total　power　consumption.　Additionally,　the　VFASHP　operates　efficiently,　achieving　an　average　COP　of　3.59.