To　address　the　challenges　of　＇pseudo-peak＇　misjudgment　rates　in　broadband　impedance　spectroscopy-based　cable　fault　localization,　this　article　proposes　a　dual-stage　optimization　method　integrating　wavelet　adaptive　denoising　with　particle　swarm-genetic　hybrid　windowing.　The　study　first　introduces　a　wavelet　adaptive　denoising　algorithm　based　on　multiscale　noise　variance　dynamic　estimation,　which　combines　the　Garrote　threshold　function　to　directionally　filter　high-frequency　noise.　This　approach　preserves　impedance　mutation　features　while　significantly　enhancing　the　signal-to-noise　ratio　and　reducing　mean　square　error.　Subsequently,　a　particle　swarm　optimization-genetic　algorithm　(PSO-GA)　hybrid　intelligent　windowing　algorithm　is　designed,　dynamically　optimizing　the　Kaiser　window’s　main　lobe　width　and　sidelobe　attenuation　factor　by　integrating　the　rapid　convergence　of　particle　swarm　optimization　with　the　global　search　capability　of　genetic　algorithms,　thereby　overcoming　the　limitations　of　traditional　window　functions　in　high-frequency　resolution.　Simulations　and　experiments　demonstrate　that　the　proposed　method　improves　the　localization　peak　by　three　times　and　optimizes　localization　accuracy　to　one-third　of　the　original　in　single-defect　scenarios,　achieving　optimal　mean　square　error　and　signal-to-noise　ratio.　In　multidefect　scenarios,　all　metrics　outperform　traditional　algorithms,　significantly　enhancing　robustness　and　localization　accuracy　in　complex　noise　environments.　©　1963-2012　IEEE.