Multivision demodulation of the FBG based on a thermal-induced chirp and a shallow neural network

We present a high-precision, low-cost demodulation method for the fiber Bragg grating (FBG) using a thermal-induced chirp and a shallow neural network. The thermal-induced chirp of a semiconductor laser generates the different wavelength components in a single pulse, which will form an exponential function echo signal after being reflected by the FBG. By learning the shape of the reflected light, the back-propagation neural network can simultaneously demodulate the sensing temperature and laser power. The whole detection system has only a few basic detection devices, which makes it low cost. The experimental results show that the multivision demodulation (MVD) method can reach a high demodulation precision of 0.35°C. We believe these results indicate the MVD method is an outstanding scheme in the field of FBG interrogation.