Depth detection of void defect in sandwich-structured immersed tunnel using elastic wave and decision tree

Abstract

Void defects seriously threaten the overall force of the Sandwich-structured immersed tunnel (SSIT). Accurately identifying the location and evaluating the severity of the void defect through non-destructive testing methods to determine the overall structural health status can provide meaningful information for the grouting reinforcement repair of the immersed tunnel structure. This research has developed a method to identify void defects in SSIT by combining impact elastic wave technology with machine learning algorithms. Eleven eigenvalues including the waveform characteristics, frequency spectrum characteristics and structural location attributes of the impulse response waveform were integrated, and the different characteristics of the void defect area and the dense area were analyzed. Based on the full-size model experimental data, a sample database was established for model training and verification. The average accuracy of void-defect identification was 90.83%. The results show that this method has great potential for the detection and evaluation of void defects in SSIT structures.