基于同步压缩变换的电缆缺陷定位研究

Research on Cable Defect Location Based on SST Transform

由于采集信号为入射波与多反射波的叠加,且多分量信号之间存在耦合不服从线性叠加原理,导致以采用魏格纳分布(Wigner-Ville distribution,WVD)为核函数的传统时频域反射法(time-frequency domain reflection,TFDR)存在严重的交叉项干扰,影响对缺陷的定位和判断。针对这一问题,本文提出一种基于同步压缩变换的电缆缺陷定位方法,该方法采用同步压缩变换(synchrosqueezing transform,SST)替代WVD以消除交叉项干扰,并以高分辨率定位电缆缺陷。同时分析参考信号相关参数,针对不同长度电缆测试,提出参考信号设计原则。该方法的主要步骤包括:通过采用TFDR法获取时域定位信号;然后采用SST求取压缩和重排后的时频谱图,以获取更精确的时频表达并消除交叉项干扰。最后,搭建了10kV交联聚乙烯电缆仿真模型进行研究,并选取了长度为500 m含中间接头的XLPE电力电缆和某地约32km海底电缆对算法进行实验验证。仿真及实验结果表明:SST算法对信号进行重排后能有效消交叉相的干扰,提高了时频域反射法定位精度。

The time-domain positioning signal collected by the traditional time-frequency domain reflection method(TFDR)is an input reflection superposition signal.Therefore,when calculating the time-frequency energy distribution of the signal using the Wigner distribution(WVD),cross-term interference occurs due to the coupling between multi-component signals.To address this issue,synchrosqueezing transform(SST)is used to compress and rearrange time-frequency coefficients,achieving high-resolution expression of complex multi-component signals,eliminating cross-term interference,and achieving high spatial resolution in the localization spectrum.Simultaneously analyze the parameters related to the reference signal and propose design principles for reference signals for testing different cable lengths.Firstly,the time-domain positioning signal is obtained by using the TFDR method.Then,SST is used to obtain the compressed and rearranged time-frequency spectrum,obtaining a more accurate time-frequency expression and eliminating cross-term interference.A 10kV cross-linked polyethylene cable simulation model was built for research,and a 500m long XLPE power cable with intermediate joints was selected for experimental verification of the algorithm.And conduct on-site testing and verification on a 32-kilometer submarine cable in a certain location.The simulation and experimental results show that the SST algorithm can effectively eliminate unwanted interference and improve the positioning accuracy of the time-frequency domain reflection method after signal rearrangement.