OPGW positioning and early warning method based on a Brillouin distributed optical fiber sensor and machine learning

Abstract

A method of optical fiber composite overhead ground wire (OPGW) positioning based on a Brillouin distributed optical fiber sensor and machine learning is proposed. A distributed Brillouin optical time-domain reflectometry (BOTDR) and Brillouin optical time-domain analyzer (BOTDA) are designed, where the ranges of BOTDR and the BOTDA are 110 km and 125 km, respectively. An unsupervised machine learning method density-based spatial clustering of applications with noise (DBSCAN) is proposed to automatically identify the splicing point based on the Brillouin frequency shift (BFS) difference of adjacent sections. An adaptive parameter selection method based on k-distance is adapted to overcome the parameter sensitivity. The validity of the proposed DBSCAN algorithm is greater than 96%, which is evaluated by three commonly external validation indices with five typical BFS curves. According to the clustering results of different fiber cores and the tower schedule of the OPGW, the connecting towers are distinguished, which is proved as a 100% recognition rate. According to the identification results of different fiber cores of both the OPGW cables and tower schedule, the connecting towers can be distinguished, and the distributed strain information is extracted directly from the BFS to strain. The abnormal region is positioned and warned according to the distributed strain measurements. The method proposed herein significantly improves the efficiency of fault positioning and early warning, which means a higher operational reliability of the OPGW cables.