A machine learning algorithm for event recognition and classification in φ-OTDR system based on lenet neural network

The full English name of OTDR is Optical Time Domain Reflectometer. The main principle of its work is the use of light in the optical fiber transmission of Rayleigh scattering and Fresnel reflection generated by the backscattering of precision photoelectric integration instrument. In recent years, deep learning methods for classifying events become more and more popular on account of its importance. But the conventional classifiers perform not that well in accuracy. In this paper, we show how conventional machine learning algorithms performs in our dataset and compare them to our new network basing on Lenet, which aims at increasing the accuracy of classification in ϕ-OTDR systems by introducing the residual module on Lenet.