In　the　process　of　grasping　workpieces,　industrial　robots　often　face　the　contradiction　of　excessive　gripping　force　causing　damage　to　the　workpiece,　and　too　little　gripping　force　leading　to　slippage.　To　address　this　issue,　a　rapid　sliding　detection　method　was　proposed　using　polyvinylidene　fluoride　(　PVDF)　piezoelectric　sensors　as　tactile　perception　elements.　First,　the　sensor　signal　was　decomposed　and　reconstructed　using　the　variational　mode　decomposition　(VMD)　optimized　by　the　Archimedes　optimization　algorithm　(　AOA)　to　reduce　noise　interference.　Next,　the　time-frequency　domain　features　of　the　signal　were　extracted　to　construct　the　signal　feature　set.　Finally,　the　dung　beetle　optimization　(DBO)　algorithm　was　used　to　optimize　the　selection　of　parameters　for　long　short-term　memory　networks　(LSTM).　The　optimized　parameters　obtained　from　DBO　along　with　the　signal　feature　set　were　applied　to　construct　the　sliding　detection　recognition　model.　The　proposed　sliding　detection　method　was　applied　to　an　experiment　involving　electric　gripper　grasping.　Results　demonstrate　precise　and　rapid　recognition　of　contact　status,　achieving　100%　accuracy　with　recognition　times　under　20　ms.　Based　on　the　recognition　results,　the　gripping　force　of　the　electric　gripper　can　be　adjusted　in　realtime.　©　2024　Chinese　Vibration　Engineering　Society.　All　rights　reserved.