Internal defect characterization of bridge cables based on Terahertz time-domain spectroscopy and deep learning

Cables are one of the most critical load-bearing components in cable-supported bridges. The condition of the cables is essential to the scientific management and safe operation of bridges. In contrast to the surface defect, the internal defect of cables is hidden, and the damage caused by it is sudden and difficult to predict, which remains a critical and challenging aspect of bridge maintenance. The commonly used internal defect detection methods, such as the magnetic leakage method and the magnetostrictive guided wave method, have difficulties in accurately locating defects. It is desirable to develop a practical nondestructive testing method for bridge cables. Therefore, this study proposes a method based on terahertz time-domain spectroscopy (THz-TDS) technology and deep learning (DL) algorithms. Firstly, the propagation theory of terahertz waves in the cable is introduced. Then, the methodology of cable defect detection based on THz-TDS and DL is proposed, including the measurement of optical properties using the transmission mode and the detection of sheath thickness and internal corrosion using the reflection mode. Finally, a series of experiments are conducted to verify the proposed method. The results indicate that the proposed method can effectively identify and locate internal corrosion in the cable. In addition, the proposed method can measure the thickness of the sheath at any position of the cable with high accuracy, showing the potential to rapidly evaluate the uniformity of the sheath thickness. •A non-destructive method to measure cable sheath thickness is proposed.•THz-TDS technology is introduced for detecting internal cable corrosion for the first time.•Corrosion defects are identified by classifying terahertz signals.