Time-frequency　domain　reflectometry　is　one　of　the　promising　methods　for　locating　latent　cable　defects　recently.　Nevertheless,　the　traditional　Wigner-Ville　distribution,　which　is　used　to　convert　the　signals　into　the　time-frequency　domain,　can　lead　to　cross-term　interference　problems　for　multi-component　signals.　To　avoid　the　influence　of　the　cross-terms　on　the　positioning　curve,　the　Gaussian　window　pseudo　Wigner-Ville　distribution　is　utilized　to　suppress　the　cross-terms.　In　the　simulation　model,　the　amplitude　of　the　cross-terms　is　greatly　reduced　without　changing　the　time-frequency　center　of　the　multi-component　signal,　which　verifies　the　ability　of　the　distribution.　In　experiments,　a　160　m　10kV　cable　was　used　as　the　test　object,　and　four　accelerated　thermal　aging　defects　which　were　heated　at　135　°C　for　different　days　were　set　at　63　m　of　the　cable　to　simulate　the　latent　defects.　The　results　show　that　the　peaks　of　the　positioning　curves　at　the　defects　based　on　the　proposed　method　in　this　paper　are　much　higher　than　that　based　on　the　traditional　one.　And　the　method　is　expected　to　be　applied　in　the　power　systems　in　the　future.　©　2022　IEEE.