Unsupervised learning method for events identification in φ-OTDR

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