High-Precision and Fast BOTDA Sensing Based on Super-Resolution Reconstruction Assistance

Abstract

The sensing performance of the Brillouin optical time-domain analysis (BOTDA) is typically limited by the frequency sweep interval, average times of time-domain trace, and the accuracy and efficiency of Brillouin frequency shift (BFS) extraction. To enhance the accuracy and response speed of BOTDA without increasing system complexity, we proposed a BFS extraction method (SuperBFSNet) with super-resolution (SR) reconstruction assistance. This method combines SR reconstruction technology to rapidly and accurately extract BFS from low-resolution Brillouin gain spectra (BGSLR) measured under a large sweep interval. The accuracy and robustness of this method under different measurement conditions were experimentally evaluated and compared with traditional Lorentz curve fitting (LCF) and reconstruction fitting based on artificial neural network (ANN) methods. Experimental results show that when the frequency sweep interval is increased by a factor of 10, SuperBFSNet exhibits smaller measurement deviations and higher stability over a wider temperature range, with a temperature linear fitting determination coefficient of 0.9984, without sacrificing spatial resolution accuracy. Furthermore, when the number of averages is greater than 50, the average BFS uncertainty at the end of the optical fiber is 0.48 MHz. This represents a 48% improvement compared to LCF extraction of 1-MHz measured Brillouin gain spectrum (BGS), with an 85% reduction in data processing time. Simultaneously, the system measurement time and data volume are reduced to 1/10.