An Improved FDR Defect Localization Method Based on K-Means Clustering and Fractional Fourier Transform

In recent years, frequency domain reflectometry (FDR) based on fast Fourier transform (FFT) has been widely used in the defect detection of cables and has been proven to be a reliable and effective detection method in many field cases. However, due to the dispersion effect, the frequency of the reflected wave varies with the adoption time, which leads to the generation of multiple reflected waves and the existence of pseudo-peaks in the localization map of defects, which may cause misjudgment of defects. The noise in the measurement environment and the noise generated by the attenuation of the test signal to distortion exacerbate the difficulty of defect detection. Therefore, to suppress the effects of pseudo-peaks and noise, this paper proposes an improved FDR defect localization method based on K-mean clustering and fractional Fourier transform (FrFT). The K-mean clustering can effectively improve the signal-to-noise ratio of the reflection coefficient spectrum, FrFT not only has the properties of traditional Fourier transform but also suppresses the pseudo-peaks generated by multiple reflections and spectral leakage. Compared with the traditional FFT, the fractional-order Fourier transform can better describe the time-frequency local characteristics of the signal and is suitable for dealing with non-smooth signals such as linear FM signals. The effectiveness and feasibility of the proposed method are verified through experiments and simulations, and the localization accuracy is improved by 62.5%.