Evaluation of welding damage in welded tubular steel structures using guided waves and a probability-based imaging approach

Welded tubular steel structures (WTSSs) are widely used in various engineering sectors, serving as major frameworks for many mechanical systems. There has been increasing awareness of introducing effective damage identification and up-to-the-minute health surveillance to WTSSs, so as to enhance structural reliability and integrity. In this study, propagation of guided waves (GWs) in a WTSS of rectangular cross-section, a true-scale model of a train bogie frame segment, was investigated using the finite element method (FEM) and experimental analysis with the purpose of evaluating welding damage in the WTSS. An active piezoelectric sensor network was designed and surface-bonded on the WTSS, to activate and collect GWs. Characteristics of GWs at different excitation frequencies were explored. A signal feature, termed 'time of maximal difference' (ToMD) in this study, was extracted from captured GW signals, based on which a concept, damage presence probability (DPP), was established. With ToMD and DPP, a probability-based damage imaging approach was developed. To enhance robustness of the approach to measurement noise and uncertainties, a two-level image fusion scheme was further proposed. As validation, the approach was employed to predict presence and location of slot-like damage in the welding zone of a WTSS. Identification results have demonstrated the effectiveness of the developed approach for identifying damage in WTSSs and its large potential for real-time health monitoring of WTSSs.