The　integration　of　reconfigurable　intelligent　surfaces　(RIS)　into　Unmanned　Aerial　Vehicles　(UAVs)　can　be　deployed　to　provide　ubiquitous　communication　services　in　communication-impaired　areas,　but　the　limited　on-board　battery　capacity　of　the　UAVs　restricts　their　endurance.　The　operating　time　of　the　UAV　can　be　extended　by　dividing　the　metasurface　elements　on　the　RIS,　but　channel　variations　due　to　the　constant　movement　of　pedestrians　present　a　significant　challenge　to　efficient　resource　allocation.　In　this　paper,　a　novel　energy　harvesting　(EH)　scheme,　called　the　UAV-EH,　is　developed　based　on　the　UAV-RIS　system.　The　proposed　UAV-EH　scheme　extends　the　Synchronized　Wireless　Information　and　Power　Transfer　(SWIPT)　system　with　simultaneous　signal　transmission　and　energy　harvesting　on　the　RIS　metasurface.　And　we　develop　a　robust　proximal　policy　optimization　(PPO)　algorithm　combined　with　cropping　the　action　space　to　assign　RIS　metasurface　elements　as　a　way　to　improve　the　UAV-EH　scheme　to　ensure　quality　of　service　(QoS)　in　dynamic　wireless　environments.　The　simulation　results　demonstrate　the　effectiveness　of　our　proposed　UAV-EH　scheme　based　on　PPO,　with　average　energy　harvesting　rates　of　65.7%　for　single-user　terminals　and　65.6%　for　multi-user　terminals.　These　rates　are　very　close　to　the　performance　of　exhaustive　search　algorithms　and　outperform　all　other　schemes.　　©　2024　IEEE.