Artificial neural network compensation for an optical fiber temperature sensing system based on optical time offset.

Through analyzing the optical time offset in optical fiber, the paper presented a new method that using optical fiber sensing technology for monitoring the temperature. Measurement experiments of monitoring temperature were carried out. Through the analysis of experimental data, the results indicated that the temperature variations could be detected by the optical fiber sensor based on optical time offset effectively, and the temperature sensitivity of the optical fiber sensing system is about 2.46mV/°C. Owing to the temperature sensor was affected by the variations and noises seriously, the artificial neural network (ANN) was introduced to the temperature sensor based on optical time offset as a compensation method, experimental results and numerical simulation indicated that the optical fiber temperature sensor with artificial neural network could enhance the nonlinear correction for sensor and decrease the interference effectively. Simultaneously, by way of the contrast test of the different neural networks, experimental results showed that the linearity of full scope could reach 0.11% for temperature sensor based on BRF network, and concluded the BRF network was more suitable for compensation of optical temperature sensor based on optical time offset.