Large Dynamic Range and Anti-Fading Phase-Sensitive OTDR Using 2-D Phase Unwrapping Via Neural Network.

Fei Peng,Xuanyu Zheng,Qiang Miao

DOI: https://doi.org/10.1109/tim.2023.3308239

IF: 5.6

2023-01-01

IEEE Transactions on Instrumentation and Measurement

Abstract:Optical fiber distributed acoustic sensing (DAS) instrument enables distributed acoustic measurements, garnering significant attention due to its high sensitivity, superior spatial resolution, and long sensing range. However, as one of the most widely adopted DAS technologies, phase-sensitive optical time-domain reflectometry (phi-OTDR) faces three key challenges in the phase demodulation of Rayleigh backscattering: 1) the original phase unwrapping (OPU) algorithm cannot recover strong acoustic signals when the phase change between the two adjacent sample time exceeds p rad; 2) when the fading occurs, even if the phase change generated by a strong acoustic wave does not exceed p rad,phi-OTDR still frequently fails to recover stronger acoustic signals with high fidelity; and 3) fading prevents phi-OTDR from recovering weak acoustic signals. This article presents a novel approach to improve both the dynamic range and anti-fading performance of phi-OTDR by utilizing a conventional phi-OTDR structure and a convolutional neural network. We trained the model by establishing a mapping relationship between the wrapped phase with noise and the real phase of multichannels. The trained model can improve both the dynamic range and anti-fading performance of phi-OTDR significantly, which has been demonstrated in 6.2-km fiber: the maximum measured strain achieved is increased from 81.1 to 365.9 ne, the self-noise is suppressed from -64 to -109 dB rad(2)/Hz, and the dynamic range reaches 122.6 dB, which is 42.8 dB greater than that of OPU.