Defect Identification for Auxiliary Power Cable in EMUs Based on TFDR and GAF-DCN

As time goes on, the auxiliary power cable of auxiliary power system in electric multiple unit (EMU) may have defects that compromise its normal operation. Although time-frequency domain reflectometry (TFDR) is capable of defect localization, it has not been utilized for identifying the types of defects. To further identify the types of defects during the localization process and enhance the efficiency of cable maintenance and repair, this article introduces a recognition approach based on a Gramian angular field deep convolutional network (GAF-DCN). First, a defect detection platform was established in the laboratory and defect cable samples were prepared. The TFDR was used to obtain the reflection signal. Then, the GAF was used to encode the reflection signal into images and DCN was used for defect-type recognition. The recognition results indicate that the proposed method can achieve an accuracy of approximately 90%. Finally, method validation was conducted in both laboratory environment and actual EMU interior, indicating that the GAF-DCN can effectively identify the types of defect.