By　performing　K-Means++　clustering　on　the　samples,　the　sample　data　set　can　obtain　a　better　initial　clustering　center,　and　the　generation　process　of　this　center　is　transformed　using　the　method　of　differential　privacy,　and　the　Gaussian　kernel　is　used　in　the　process　of　updating　the　clustering　center.　The　function　performs　privacy　allocation　and　adds　noise,　and　finally　uses　the　particle　swarm　optimization　algorithm　to　optimize　and　the　experimental　results　show　that　the　algorithm　has　achieved　good　prediction　results,　and　the　indicators　are　equal　to　the　original　algorithm,　so　that　the　algorithm　can　protect　the　user＇s　privacy　while　ensuring　the　accurate　clustering　center.　©　2023　IEEE.