Deep neural networks assisted BOTDA for simultaneous temperature and strain measurement with enhanced accuracy

Abstract

Simultaneous temperature and strain measurement with enhanced accuracy by using Deep Neural Networks (DNN) assisted Brillouin optical time domain analyzer (BOTDA) has been demonstrated. After trained by using combined ideal clean and noisy BGSs, the DNN is applied to extract both the temperature and strain directly from the measured double-peak BGS in large-effective-area fiber (LEAF). Both simulated and experimental data under different temperature and strain conditions have been used to verify the reliability of DNN-based simultaneous temperature and strain measurement, and demonstrate its advantages over BOTDA with the conventional equations solving method. Avoiding the small matrix determinant-induced large error, our DNN approach significantly improves the measurement accuracy. For a 24-km LEAF sensing fiber with a spatial resolution of 2m, the root mean square error (RMSE) and standard deviation (SD) of the measured temperature/strain by using DNN are improved to be 4.2 degrees C/134.2 mu epsilon and 2.4 degrees C/66.2 mu epsilon, respectively, which are much lower than the RMSE of 30.100710.2 mu epsilon and SD of 19.4 degrees C/529.1 mu epsilon for the conventional equations solving method. Moreover, the temperature and strain extraction by DNN from 600,000 BGSs along 24-km LEAF requires only 1.6s, which is much shorter than that of 5656.3s by the conventional equations solving method. The enhanced accuracy and fast processing speed make the DNN approach a practical way of achieving simultaneous temperature and strain measurement by the conventional BOTDA system without adding system complexity.