Machine learning applied to BOTDR optical fibre distributed sensing in a controlled environment

Abstract

We present a study on the application of machine learning to optical fibre distributed sensing, with data recovered using a state-of-the-art, commercial BOTDR distributed sensing system, to extract temperature information from single-mode optical fibre over a 40-km distance. The application is for power line monitoring of underground cables that are collocated with optical fibres that form part of the Electricity Authority of Cyprus' island wide power distribution networks. The existing optical fibre infrastructure acts as the sensing element, monitoring temperature changes when in close proximity to the power lines. The initial training measurements for the machine learning algorithm were recorded in a laboratory setting using temperature and humidity-controlled elements, with sections of fibre spliced to underground fibre cables subjected to temperature excursions. A machine learning approach was implemented for the prediction task of finding points that are likely to get damaged, mimicking the behavior of power cable joints that are prone to failure, along with general monitoring for unusual behavior and potential cable fault conditions; the task is a binary classification one. Labels "0/1"were assigned to the BOTDR measurements, with "1"corresponding to data points in space and time for which the signal showcased a problematic scenario, such as the collocated fibre's temperature rising to dangerously high values, and "0"to the rest. The algorithm's base is a variation of the state-of-the-art transformer architecture, which depends solely on attention mechanisms. The training was undertaken on the laboratory data and re-training is done periodically with new field measurements. The completion of the training phase shows the potential of the algorithm to predict spatiotemporally problematic points, using the temperature measurements of the collocated fibre; this will be extended to BOTDR data taken in the field.