Regarding　the　problem　that　fuzzy　c-means　algorithm　(FCM)　was　sensitive　to　the　initial　value　and　converging　to　the　local　infinitesimal　point　easily,　applies　genetic　algorithm　to　optimization　of　the　FCM　algorithm.　Firstly,　the　results　of　FCM　will　be　sent　to　the　genetic　algorithm　for　optimization,　then　the　new　results　again　used　in　FCM　to　obtain　the　most　advantage　of　the　overall　situation.　The　experimental　result　shows　that　the　algorithm　can　effectively　detect　anomaly　intrusions　behavior　of　special　target　and　be　better　than　FCM　algorithm,　and　have　a　strong　global　optimization　and　faster　convergence　speed.