High-dimensional　data　clustering　is　of　great　importance　in　the　big　data　era.　Multi-objective　evolutionary　soft　subspace　clustering　(SSC)　algorithms　have　shown　promise　in　handling　such　datasets,　but　the　objective　functions　and　local　search　strategies　used　have　not　yet　been　well　investigated.　To　consider　these　issues,　this　paper　proposes　an　improved　multi-objective　evolutionary　approach　with　new　objective　function　and　local　search　operator　for　clustering　high-dimensional　data.　First,　a　new　objective　function　is　provided,　which　optimizes　the　clustering　validity　indexes　and　additional　item　simultaneously　to　overcome　the　difficulty　of　coefficient　settings　in　the　objective　functions　of　existing　SSC　approaches.　Second,　an　improved　local　search　operator　is　introduced,　which　updates　the　weights　of　features　by　considering　both　the　within-class　compactness　and　between-class　separation　to　capture　a　more　comprehensive　data　structure.　An　experimental　study　with　comparison　with　state-of-the-art　SSC　methods　demonstrates　the　efficiency　of　the　proposed　approach.