The　SiC/SiO2　interface　state　is　one　of　the　main　factors　that　limit　the　performance　and　reliability　of　the　SiC　metal-oxide-semiconductor　field-effect　transistor　(MOSFET).　In　this　article,　we　use　a　Bayesian　deconvolution　algorithm　to　optimize　trap　feature　extraction　based　on　the　transient　current　method　and　improve　the　trap　extraction　accuracy.　Using　this　method,　we　study　the　trap　capture　mechanism　in　SiC　MOSFETs　and　mainly　characterize　the　trap　position,　the　trap　energy　level,　and　the　capture　time　constant.　The　results　obtained　show　that　there　are　three　different　types　of　traps　and　defects,　two　of　which　are　SiC　interface　traps　at　the　gate-source　and　gate-drain　interfaces,　with　activation　energies　of　0.089　and　0.035　eV,　respectively,　and　the　third　trap　type　is　an　oxide　trap,　and　its　time　constant　does　not　vary　with　temperature.　The　characterization　results　are　verified　via　deep-level　transient　spectroscopy,　and　the　results　show　reasonable　agreement　with　those　obtained　by　the　method　proposed　in　this　article.　This　method　can　be　combined　with　electrical　stress　testing　in　long-term　reliability　research　to　realize　nondestructive　characterization　of　the　defects　of　SiC　MOSFETs.