In　recent　years,　with　the　development　of　computer　technology,　EEG　emotion　recognition　has　been　paid　much　attention　in　the　field　of　emotional　computing　and　has　become　a　hotspot.　In　the　practical　application　of　EEC　emotion　recognition,　it　is　not　only　required　to　ensure　the　accuracy　rate,　but　also　the　operation　efficiency.　Based　on　this,　we　propose　a　ReGA　algorithm　to　select　the　EEG　characteristics.　In　the　ReGA　algorithm,　the　ReliefF　algorithm　is　used　to　calculate　the　feature　weight,　and　the　heuristic　information　is　provided　for　the　population　initialization　of　the　encapsulation　stage　genetic　algorithm,　so　that　the　initial　population　contains　a　good　starting　point,　so　the　genetic　algorithm　can　adopt　less　evolution　Algebra　and　small-scale　populations　to　find　a　better　subset　of　features.　Therefore,　the　ReGA　algorithm　can　avoid　the　danger　of　ReliefF　to　remove　important　features,　but　also　reduce　the　probability　of　genetic　algorithm　to　adapt　and　improve　the　efficiency　of　computing.