Abstract: It is difficult to accurately monitor the threedimensional deformation of joints in underwater tunnels based on traditional methods. A new threedimensional deformation sensing method for tunnel joints is proposed based on distributed fiber optic strain sensing technology and artificial intelligence machine learning algorithm. Firstly, a monitoring method of joint deformation of tunnel is designed by using distributed optical fiber strain sensor. According to the theoretical relationship between the optical fiber strain and the joint deformation, a threedimensional deformation model of tunnel joints is established, and 12 sets of data are obtained. There are 4 000 data of tunnel joint deformation and optical fiber strain in each set. Then, a new twolevel progressive classification algorithm based on machine learning methods, including decision tree, random forest, and support vector machine, is proposed to accurately calculate the threedimensional deformation of tunnel joints. The accuracy of tunnel joint deformation can reach 0.1 mm. It is found that the support vector machine has higher accuracy and stability for threedimensional deformation of tunnel joints compared with the machine learning algorithms of decision tree and random forest.

Key words:  underwater tunnel, joint deformation, distributed optical fiber sensing, strain, machine learning