With　the　increasing　proportion　of　photovoltaic　power　generation　in　the　power　system　year　by　year,　the　prediction　of　photovoltaic　power　generation　power　is　increasingly　valued.　This　article　proposed　a　photovoltaic　power　generation　power　prediction　method　based　on　the　RF-XGBoost　model　to　address　the　issue　of　unstable　output　power　of　photovoltaic　power　generation　systems　when　integrated　into　the　power　grid,　using　meteorological　data　for　prediction.　This　method　combined　the　Random　Forest　algorithm　(RF)　and　XGBoost　algorithm,　and　fused　their　prediction　results　by　weighted　average,　giving　full　play　to　the　advantages　of　the　two　models,　thus　improving　the　accuracy　and　reliability　of　photovoltaic　power　prediction.　Through　the　training　and　verification　of　historical　data,　the　experimental　results　have　shown　that　the　RF-XGBoost　model　is　superior　to　the　single　XGBoost　model,　Random　Forest　model　and　RF-SVR　model,　and　has　achieved　remarkable　results　in　prediction　accuracy　and　performance　advantages.　This　method　has　provided　reliable　decision-making　support　for　the　optimized　operation　of　photovoltaic　power　generation　systems　and　grid　management,　and　had　an　important　contribution　to　promoting　the　large-scale　application　of　photovoltaic　power　generation.　©　2023　IEEE.