Event Detection for Distributed Acoustic Sensing: Combining Knowledge-Based, Classical Machine Learning, and Deep Learning Approaches.

Mugdim Bublin

doi:10.3390/s21227527

Abstract

Distributed Acoustic Sensing (DAS) is a promising new technology for pipeline monitoring and protection. However, a big challenge is distinguishing between relevant events, like intrusion by an excavator near the pipeline, and interference, like land machines. This paper investigates whether it is possible to achieve adequate detection accuracy with classic machine learning algorithms using simulations and real system implementation. Then, we compare classical machine learning with a deep learning approach and analyze the advantages and disadvantages of both approaches. Although acceptable performance can be achieved with both approaches, preliminary results show that deep learning is the more promising approach, eliminating the need for laborious feature extraction and offering a six times lower event detection delay and twelve times lower execution time. However, we achieved the best results by combining deep learning with the knowledge-based and classical machine learning approaches. At the end of this manuscript, we propose general guidelines for efficient system design combining knowledge-based, classical machine learning, and deep learning approaches.

Highlights

Pipeline monitoring and protection is often performed by expensive and laborious physical inspection methods such as helicopter or vehicle patrols [2]
We describe an application of the classic machine learning (ML) approach in event detection with Distributed acoustic-optical sensors (DAS) systems
The task of Feature Extraction is to extract from the raw data the features that are representative for the events of interest with enough discriminative power for subsequent classification by machine learning algorithms

Summary

Introduction

Pipeline monitoring and protection is often performed by expensive and laborious physical inspection methods such as helicopter or vehicle patrols [2]. Distributed acoustic-optical sensors (DAS) are a comparatively new approach (see Figure 1) Such systems permit the supervision of long-distance pipelines using fiber optics [3,4,5,6,7,8,9,10]. It was demonstrated that the event of a person climbing the fence can be distinguished from the wind-induced fence vibration, and the resulting false alarm rate was around one per month during a 1-year test period. The details of optical setup and pulse forming as well as signal processing for the detection and location of the intruder from the backscattered signal of an interrogating pulse are described, but the classification of the event is not disclosed. Harman discloses a method and apparatus in PCT patent application WO2013/185208 [10] for short-range perimeter surveillance with two back-to-back Michelson interferometers using a cable comprising of four optical fibers

Methods

Results

Conclusion