Abstract: With the increase of the service life,various kinds of latent defects gradually appear in cables. It is of great significance to diagnose defects before they cause faults.Existing cable detection methods such as dielectric loss measurement and partial discharge detection have their own scope of application and cannot detect and locate defects such as aging and partial breakage of cables before they cause faults.This paper proposes a non-destructive detection method based on the impedance spectroscopy,which can effectively locate early cable defects and avoid evolving to produce cable fault accidents. The paper first establishes the numerical model of cable distribution parameters,studies the characteristic law of cable impedance spectrum, comparing and analyzing the influence of the appearance of latent defects with different ageing degrees on the cable impedance spectrum,and analyzes the frequency domain impedance spectrum characteristics of different latent defects. Based on this, a Discrete Fourier Transform is applied to the impedance spectrum of the cable containing defects,and the results of the integral transform are processed by the Chebyshev Window function to achieve the accurate localization of the latent defects in multiple segments. It is shown in experiments and simulations that the method could locate latent defects in multiple sections of cable,and the window function algorithm could significantly suppress the noise and improve the location accuracy,which provides a new technical means for the improvement of cable fault location technology.