Comparison of non-destructive testing methods of bolted joint status in steel structures

This article analyzes the commonly used non-destructive testing methods of bolted joints. The contact method has high recognition accuracy for preload detection but is greatly affected by the coupling between sensors and interfaces. The non-contact method is convenient for collecting signals, but its recognition accuracy is greatly affected by environmental noise. For stress identification of steel plates, the identification accuracy is limited by the interface coupling situation and the shallow stress can only be identified. The evaluation of bolted joint status is mainly characterized by interface stiffness, parametric research should be conducted to explain the mapping mechanism between stress waves and interface stiffness. Machine learning can help improve the accuracy and efficiency of damage detection, but the end-to-end recognition model does not have physical significance. Therefore, the hardware and lightweight, intelligent recognition algorithms should be developed to improve the computational efficiency of the model and physical interpretability.