Acoustic Imaging with Distributed Acoustic Sensing Data Using Delay-and-Sum Beamforming in the Time Domain

Abstract

Distributed Acoustic Sensing (DAS) is an emerging technology that uses an optical fibre as a distributed vibration sensor. Strain variations in the fibre are measured by phase changes in the reflections captured by an interrogator at one end of the sensing fibre, thanks to Rayleigh back-scattering. These data enable the implementation of array processing algorithms to identify acoustic sources both in the near and far field. However, array processing for DAS data has certain limitations, as the acoustic field is indirectly sampled as longitudinal tension variations in the fibre, which may also experience coupling issues in some sections, complicating the direct application of array processing principles. Beamforming is an array processing technique that focuses the listening capabilities of an array on a specific section of space by introducing phase shifts in the measurements from each sensor. The performance of delay-and-sum beamforming applied to DAS measurements is evaluated in a near-field scenario using a public dataset. In this case, a 1500 by 500 metre fibre is interrogated by a DAS sensor to measure controlled acoustic vibrations generated by a truck at various positions, with 55 positions arbitrarily selected for analysis. During the experiment, the fibre also measures vibrations from a distant earthquake, enabling evaluation in a far field scenario. To overcome the limitations of array signal processing applied to Distributed Acoustic Sensing, a channel selection method based on a similarity score, calculated from phase correlation between channels, is implemented, capitalising on the high channel availability in DAS. It is demonstrated that, in most cases, a limited selection of channels improves the estimation of acoustic sources com pared to using all available channels. Source localisation is achieved independently by triangulation and beamforming-based acoustic imaging for the near field, with the latter method being the main focus of this thesis, providing the potential for the estimation of multiple sources simultaneously. For the far field, acoustic imaging is employed across a range of propagation speeds and angles to estimate the direction of arrival of an earthquake.