Accurate　prediction　of　the　remaining　useful　life　(RUL)　of　rolling　bearings　in　mechanical　equipment　is　crucial　for　ensuring　the　reliable　operation　of　equipment　and　implementing　effective　maintenance　measures.　It　also　plays　a　key　role　in　safeguarding　personnel　and　property　by　reducing　the　risk　of　failures.　Recently,　converting　monitoring　data　into　a　graph　structure　to　capture　the　mutual　influence　between　samples　has　emerged　as　an　innovative　approach　in　RUL　prediction.　However,　existing　methods　cannot　effectively　extract　features　from　graph-structured　data　with　varying　receptive　fields　and　establish　strong　dependencies　between　nodes.　This　research　proposes　a　novel　bearing　RUL　prediction　model　based　on　the　multi-region　hypergraph　self-attention　network　(M-HGSAN)　to　address　these　challenges.　Firstly,　the　original　data　is　concatenated　and　resampled　using　the　sliding　window　with　width　L,　and　the　two-dimensional　sample　set　is　constructed　by　time　domain　feature　and　frequency　domain　feature,　which　enriches　the　diversity　of　samples.　The　multi-scale　synchronous　semi-shrink　attention　network　(MSSSAN)　is　used　to　obtain　different　channel　features　and　multi-region　features　from　different　receptive　fields,　which　enhances　the　dependence　between　features.　Secondly,　a　hypergraph　selfattention　network　(HGSAN)　is　designed,　which　combines　the　advantages　of　a　hypergraph　neural　network　(HGNN)　and　a　self-attention　mechanism.　Obtain　the　ability　to　learn　higher-order　correlation　and　key　features　between　nodes.　In　addition,　the　data　is　fed　into　residual　stacked　gated　recurrent　units　(RSGRU)　and　fully　connected　(FC)　layers　to　capture　the　nodes＇　temporal　sequence　features　and　predict　the　bearings＇　RUL.　Finally,　model　interpretability　experiments　are　carried　out　with　XAI　technology　to　help　us　understand　the　influence　of　each　feature　on　RUL.　Experimental　results　demonstrate　the　effectiveness　of　the　M-HGSAN　model,　highlighting　its　potential　to　significantly　enhance　predictive　maintenance　strategies　in　industrial　applications,　thereby　improving　equipment　reliability　and　safety.