Impact-Based Feature Extraction Utilizing Differential Signals of Phase-Sensitive OTDR

Abstract

Traditional vibration pattern recognition of $\phi$ -OTDR system was operated by event-based feature extractors. The proposed adjacent Spearman correlation assisted algorithm can be considered as impact-based feature extractor utilizing differential signals by extracting the impact region of the perturbation without involving idle data region. This technique can utilize less number of data traces in the pattern recognition process. Classifying with the unique way of impact-based strategy for the direct detected $\phi$ -OTDR system, both the classification accuracy and processing cost have been improved irrespective of the nature of applied perturbation.