کنفرانس مهندسی برق ایران

صفحه اصلی / سی و دومین کنفرانس بین المللی مهندسی برق

Φ-OTDR Event Classification Using Machine Learning and Optical Signal Processing

نویسندگان :

Amir Babaoughli¹ Tohid Alizadeh² Seyed Sadra Kashef³

1- Urmia University 2- Urmia University 3- Urmia University

کلمات کلیدی :

OTDR،Fiber optic sensing،Event classification،Machine learning،SVM

چکیده :

Industries such as power communication networks, oil and gas pipelines are among the mother and infrastructure industries, so their continuous maintenance and monitoring are very important. Automating the monitoring of these industries increases their productivity. This work involves monitoring using phase-sensitive optical time domain reflectometry technology along with machine learning. In this method, the features are extracted and pre-processed by maximum relevance - minimum redundancy, and principal component analysis and then given to the support vector machine classifier. Maximum relevance - minimum redundancy ranks the features and principal component analysis improves the performance of support vector machine classifier by mapping the high-score features to a space with high resolution and selecting the features with high information and increases the accuracy of the system. The accuracy of the proposed method was evaluated at 95.46 percent, which significantly improves over previous methods. It is worth mentioning that the dataset used in this study is real and produced in a laboratory environment.