Research On Disease Detection Method Of Tunnel Lining Structure Based On Game Theory And Deep Neural Network

With the increasing amount of tunnel construction in China,the role of tunnels in the transportation road network is becoming more and more prominent.However,during the construction process and actual operation and maintenance of tunnels,affected by various factors such as construction quality,natural climate change,harsh service environment and complex internal and external loads,various diseases may occur in the tunnel lining structure,jeopardizing the safe operation of tunnels.Therefore,the efficient detection of tunnel lining defects is essential to improve the structural stability and ensure the safe operation of tunnels.In this paper,the main objective is to achieve accurate and efficient identification of tunnel lining structure diseases by conducting research in the following areas:(1)To address the problems of low positive sample data and low clarity of image structure in GPR image data of tunnel lining structure disease,a method of generating GPR images of tunnel lining structure disease is proposed by combining generative adversarial network with game idea.The experiments show that the method can effectively generate tunnel lining structure disease GPR images and optimize the clarity of the disease structure.(2)In response to issues such as low artificial recognition efficiency and poor accuracy,a tunnel lining structure disease recognition algorithm CNM-YOLO based on attention mechanism is proposed.Using YOLOv5 s algorithm as the benchmark,a Merge CSA feature extraction module with fused attention mechanism was constructed,and Conv Next was combined with an adaptive spatial fusion module to deeply fuse the lining disease features.Experimental results show that compared with other algorithms,this algorithm has certain advantages in the recognition of tunnel lining structure diseases.(3)Engineering case application.Based on the Chenzhai Tunnel of Lianhuo Equang Expressway Liaison Line(Xin’an-Yichuan Expressway)and a section of the Fengjialiang Tunnel in Pingli County,Ankang City,the algorithm proposed in this paper was used for detection.The algorithm detection results were compared with the results of professional engineering testing reports,and it was found that the discrimination results were basically consistent.This indicates that the method for identifying tunnel lining diseases proposed in this paper has certain practical value.