As　of　2018,　China　had　more　than　550,000　data　centers,　accounting　for　approximately　1.5%　of　the　country＇s　total　annual　electricity　consumption.　Reducing　energy　consumption　not　only　lowers　costs　but　also　aligns　with　environmental　protection　principles.　Current　research　primarily　focuses　on　IT　equipment　and　air　conditioning　and　refrigeration　systems,　as　these　components　offer　significant　potential　for　energy　savings.　Instead　of　making　adjustments　to　the　original　space　environment　and　hardware　facilities　of　the　data　center,　this　study　proposes　an　energy　efficiency　optimization　model　by　regulating　the　set　values　and　parameters　of　the　cooling　system.　This　approach　is　particularly　valuable　given　the　increasing　emphasis　on　energy　efficiency　in　data　centers.　The　cooling　system　of　a　data　center　is　highly　complex,　with　multiple　variables,　high　feature　dimension,　time-varying　working　conditions,　and　strong　uncertainty.　This　study　employs　an　artificial　neural　network　to　model　the　energy　efficiency　(PUE)　of　a　data　center,　using　data　collected　from　the　operation　of　the　cooling　system　in　a　data　center　in　southwest　China.　Subsequently,　an　evolutionary　algorithm　is　used　to　analyze　and　optimize　the　cooling　system,　controlling　the　set　values　to　enhance　the　energy　efficiency　of　the　data　center.　©　2024　IEEE.