Unsupervised learning method for events identification in φ-OTDR

Abstract

In this paper, an unsupervised-learning method for events-identi cation in φ-OTDR ber-optic distributed vibration sensor is proposed. The different vibration-events including blowing, raining, direct and indirect hitting, and noise-induced false vibration are clustered by the k-means algorithm. The equivalent classi cation accuracy of 99.4% has been obtained, compared with the actual classes of vibration-events in the experiment. With the cluster-number of 3, the maximal Calinski-Harabaz index and Silhouette coe cient are obtained as 2653 and 0.7206, respectively. It is found that our clustering method is effective for the events-identi cation of φ-OTDR without any prior labels, which provides an interesting application of unsupervised-learning in self-classi cation of vibration-events for φ-OTDR.

Development of optical fiber strain sensor system based on machine learning and polarization

“…The classification accuracy reached 99.4% compared to the actual classification. 18) Sai et al proposed collision detection and localization based on collision localization system in FBG sensor networks. According to the energy of the signal in the available frequency band, an impact localization model based on the extreme learning machine (ELM) is established with faster training speed and fewer parameters.…”