Artificial　neural　network,　as　the　basis　of　the　continuous　development　of　the　current　field　of　artificial　intelligence,　contains　many　types　of　algorithms,　but　most　of　them　contain　hidden　layer.　By　understanding　the　prediction　accuracy　of　the　previous　RBF　neural　network,　it　is　easy　to　affect　its　own　application　efficiency　and　numerical　changes　due　to　the　unsuitable　central　nodes　of　the　randomly　selected　hidden　layer.　Therefore,　under　the　background　of　new　era,　in　order　to　better　show　the　effectiveness　of　the　application　of　artificial　neural　network,　based　on　the　understanding　on　the　basis　of　application　of　RBF　neural　network,　using　the　neighbor　spread　AP　clustering　algorithm　to　carry　on　the　improvement　and　optimization,　and　after　the　new　optimization　method　and　its　modeling　process,　according　to　the　simulation　analysis　judgment　finally　optimization　method　is　scientific　and　effective.　©　2021　IEEE.