Tunnel damage identification method based on relative entropy of wavelet packet energy: An experimental verification

Ineffective methods used to detect damage in subway tunnels can result in severe safety risk. Traditional detection methods use laser scanning, image recognition, and inspection to identify damages in the inner lining of a tunnel after the subway is out of service; these methods have low efficiency and accuracy. In this study, a new method is proposed to analyze the vibration signal of a moving train based on the relative entropy of the wavelet packet energy, which can quickly identify and evaluate the damage of the subway tunnel and its auxiliary structure in real time. The test model includes a three-dimensional printed tunnel and a moving vehicle to simulate the train running in the tunnel. Wireless acceleration sensors are installed inside the vehicle and at the top of the tunnel to record the vibration signals of the car and the tunnel lining. After obtaining the acceleration signal of the vehicle, the relative entropy of energy is calculated using wavelet packet transform, in which sudden changes in entropy reflect damages. The model test results show that the vertical acceleration of the vehicle is sensitive to tunnel damage and the signal energy is mainly concentrated at 30–80 Hz and 200–400 Hz. By comparing the energy of the relative entropy between healthy and damaged signals, the location and degree of damage can be identified.