Paper Title
Incidents Management in Fiber Optic Transmission Networks by Neural Network Method

Abstract
We propose a model to systematize in real time the management of incidents in fiber optic transmission networks. Concretely, this is done in several steps: first, images of the optical infrastructure of a telecommunication operator in Senegal are collected during the different interventions of survey, deployment, maintenance or supervision of the operator's optical network. Then, the collected images of the optical infrastructure are used to train a convolutional neural network (CNN) to predict from the image received as input, if there are defects or not on the optical fiber. Finally, we will review the experimental results of such a system based on artificial intelligence (AI). This automatic system, in real time, offers a gain in efficiency and productivity in the management of incidents by allowing the operator, thanks to artificial intelligence (AI) applied to image recognition, to identify and notify the competent team of incidents such as the cutting of the optical fiber. At the same time, such a system is also very useful for the telecom regulator to be able to verify the quality of the fiber optic installations, the respect of the standards of the operators and the fiber optic access providers from the images. Keywords - Fiber Optics, Incidents, Convolutional Neural Network, Artificial Intelligence, Image Recognition, Telecommunications.