Width estimation of hidden cracks in tunnel lining based on time-frequency analysis of GPR data and back propagation neural network optimized by genetic algorithm
Lili Hou, Qian Zhang, Yanliang Du
Abstract
The width and buried depth of hidden cracks in tunnel lining are important indicators for measuring the development degree of crack propagation , and evaluating the risk of lining block falling caused by the coexisting defects with voids and cracks. This paper proposes a high-precision fitting method for the width and buried depth of hidden cracks in lining that does not rely on wave velocity. Time domain features, as well as the main frequency, 3 dB bandwidth , and 3 dB energy of the partial time energy spectrum , are used as the inputs of the back propagation (BP) neural network . The experimental results show that compared with models trained by time domain features, models trained by time domain and time-frequency domain features have better fitting accuracy and generalization ability, the estimation results of crack width and burial depth can be used for monitoring the development speed of the crack propagation in void lining.