In　the　autonomous　UAV　cruise　mission,　safety　and　reliability　are　critical　and　challenging　issues.　From　the　historical　UAV　accidents,　it　is　clear　that　ensuring　UAV　operation　safety　is　important.　In　order　to　ensure　that　UAV　can　cruise　according　to　the　preset　safe　path,　a　prediction　model　based　on　deep　neural　network　is　proposed　in　this　paper.　The　stacked　Bidirectional　and　Unidirectional　LSTM　(SBULSTM)　network　uses　the　four　positions　before　the　current　time　of　UAV　to　predict　the　position　of　the　next　time　during　cruise　operation.　UAV　is　safely　controlled　according　to　the　linear　distance　between　the　real　position　and　the　predicted　position.　When　the　linear　distance　is　not　greater　than　the　set　threshold,　UAV　can　independently　and　safely　complete　the　cruise　task.　©　2021　IEEE.Allrights　reserved