反射系数
频域
指数增长
采样(信号处理)
反射(计算机编程)
声学
频谱
光谱(功能分析)
统计
数学
物理
计算机科学
光学
数学分析
光谱密度
探测器
量子力学
程序设计语言
作者
Kuangyi Jiang,Kai Zhou,Xiang Ren,Yefei Xu
出处
期刊:Energies
[Multidisciplinary Digital Publishing Institute]
日期:2025-05-08
卷期号:18 (10): 2428-2428
被引量:1
摘要
The existing linear frequency increment cable defect detection method using frequency domain reflectometry suffers from severe pseudo-peak phenomena due to non-targeted frequency domain sampling, which interferes with diagnosis. To address this issue, this paper proposes an optimized frequency domain sampling method based on the exponential frequency increment reflection coefficient spectrum. This method optimizes the distribution of frequency domain sampling points, reducing the sampling of high-frequency noise signals, thereby effectively suppressing pseudo-peaks. Research indicates that the low-frequency band of the cable reflection coefficient spectrum contains richer information about the cable’s condition and has less noise compared to the high-frequency band. Therefore, an exponential frequency increment is used instead of the current linear frequency increment, resulting in a denser sampling in the low-frequency band and sparser sampling in the high-frequency band, better matching the information distribution characteristics of the cable reflection coefficient spectrum. To avoid spectral leakage caused by non-uniform sampling under exponential frequency increments, this method locally linearizes exponential sampling and uses interpolation to complete the overall frequency sampling rate, ensuring it meets the basic assumption of Fourier transform—uniform and equally spaced sampling signals. Finally, this method was validated on a 500 m laboratory test cable and a 2000 m operational cable. Experimental results show that this method can make the amplitude of regions other than impedance mismatch points in the positioning curve flatter and effectively suppress abnormal peak interference, significantly improving the accuracy of defect diagnosis.
科研通智能强力驱动
Strongly Powered by AbleSci AI