The　tribological　performance　and　wear　mechanism　of　polyetheretherketone　(PEEK)/17-4PH　stainless　steel,　PEEK/silicon　carbide　(SiC),　WC-6Ni　(YN6X)/SiC　and　SiC/polycrystalline　diamond　composite　(PCD)　tribopairs　sliding　in　seawater　at　different　temperatures　(ranged　from　25　degrees　C　to　70　degrees　C)　and　salinities　(20%o,　35%o　and　50%o)　was　investigated.　A　deep　neural　network　model　was　used　to　predict　the　coefficient　of　friction,　combining　a　onedimensional　CNN　and　an　LSTM.　The　experiment　results　showed　that　increasing　salinity　led　to　a　decrease　in　tribological　performance　of　the　tribopairs,　while　the　performance　effectively　improved　within　the　temperature　range　of　25-55　degrees　C.　The　CNN-LSTM　model　demonstrated　high　accuracy　in　predicting　results,　which　is　significant　for　analyzing　the　tribological　characteristics　of　tribopairs　in　seawater　hydraulic　components.