This　study　investigated　the　axial　compressive　behavior　of　UHPC-filled　glass-fiber-reinforced　polymer　pultruded　square　columns　(UFGCs)　and　UFGCs　reinforced　with　CFRP　fabric　(UFGRCCs).　The　influence　of　stirrup　type,　stirrup　pitch,　and　CFRP　fabric　width　on　the　axial　compressive　behavior　was　discussed.　The　experimental　results　demonstrated　that　placing　stirrups　with　an　appropriate　pitch　within　UHPC　can　inhibit　crack　formation　and　prevent　premature　loss　of　load-bearing　capacity.　The　use　of　CFRP　fabric　to　reinforce　the　pultruded　column　strengthens　the　confinement　effect　on　UHPC.　FE　analysis　demonstrated　that　the　diminishing　improvement　effect　of　CFRP　fabric　on　the　mechanical　properties　of　UHPC　was　observed　when　the　number　of　CFRP　fabric　layers　exceeded　7.　The　results　of　FE　parametric　expansion　analysis　were　utilized　to　establish　a　neural　network　predictive　model　for　load-bearing　capacity　(Pmax)　and　ductility　(TI).　This　model　not　only　demonstrated　outstanding　predictive　accuracy　but　also　served　as　an　objective　function　for　Pmax　and　TI.　To　maximize　the　objective　functions,　a　non-dominated　sorting　genetic　algorithm　with　an　elitist　strategy　(NSGA-II)　was　utilized　for　multi　objective　optimization.　The　optimal　reinforcement　method　for　the　CFRP　fabric　was　then　determined　using　the　technique　for　order　of　preference　by　similarity　to　the　ideal　solution　(TOPSIS).