The　critical　step　in　the　mechanical　fault　diagnoses　of　a　high-voltage　circuit　breaker　(HVCB)　is　the　extraction　of　mechanical　state　characteristics.　The　effectiveness　of　these　characteristics　is　related　to　the　accuracy　of　the　corresponding　diagnostic　results.　A　novel　method　for　the　mechanical　fault　feature　extraction　of　a　HVCB　via　complete　ensemble　empirical　mode　decomposition　with　adaptive　noise　(CEEMDAN)　and　weighted　time-frequency　entropy　is　proposed　in　this　study.　First,　the　time-frequency　decomposition　of　vibration　signals　is　performed　by　the　CEEMDAN　and　the　Hilbert　transform　to　obtain　the　instantaneous　frequency　of　intrinsic　mode　function　(IMF)　components.　Then,　the　IMFs　are　reconstructed　via　a　bandpass　filter　to　gain　a　signal　component　of　the　specified　frequency　band.　In　addition,　the　weighted　time-frequency　entropy　feature　is　extracted　from　the　energy　matrix.　Through　the　verification　of　multiple　groups　of　classifiers,　the　diagnostic　results　have　a　higher　accuracy,　which　indicates　that　the　proposed　method　can　effectively　characterize　different　mechanical　faults.　©　2019　IEEE.