Crack extension identification based on distributed fiber sensing measurement and optimized support vector regression

Abstract

Crack is one of the main causes for the failure of marine structure due to the action of waves and ocean. It is necessary to monitor the initiation and extension of the crack timely and accurately, so as to provide effective suggestion for maintenance and prolong the service life of the structure. A novel distributed optical fiber sensing tape based on optical frequency domain reflectometer (OFDR) has been proposed in previous research, which can be used to find and locate cracks. In this paper, distributed optical fiber sensing (DOFS) technology is used to obtain the strain data around the crack, and the identification of crack extension is carried out based on the strain data combining with support vector regression (SVR). The parameter set of the SVR is optimized by particle swarm optimization (PSO) to improve the accuracy of the identification model. The results show that the crack extension identification system has a root mean square error (RMSE) as low as 3.59 when no noise is added. The influence of noise is investigated by simulation. The results show that the mean relative error (MRE) is 0.066 when 5% Gaussian white noise is added, which indicates that the crack extension length can still be identified accurately under the influence of noise.