Fiber Optic Sensor Demodulation System Using Peak Wavelength Tracking Method and Neural Network Algorithm

Fiber optic sensors are used to measure various physical quantities because of their excellent characteristics. This paper proposes a fiber optic sensor demodulation system for temperature and barometric pressure measurement using an artificial neural network. The fiber optic sensor uses the structure of a fiber Bragg grating tandem Fabry-Perot interferometer, which is used to convert changes in temperature and barometric pressure in the environment into changes in the peak wavelength of the reflected spectrum. The demodulation system uses an artificial neural network to establish the nonlinear relationship between the peak spectral wavelength and the temperature and pressure to achieve the demodulation of the fiber optic sensor. An effective data augmentation method is used to train the neural network model to guarantee the demodulation performance even for small experimental data. The experimental results show that the average demodulation error of this system is 0.48 °C for temperature and 13.38kPa for barometric pressure, and the demodulation performance is better than the traditional wavelength tracking demodulation system. This paper provides a practical reference scheme for neural network application in fiber optic sensing demodulation.