FBG Micro-strain Sensor Demodulation System Based on Back Propagation Neural Network

A high-performance and low-complexity demodulation system is essential for the demodulation of FBG sensors. In this paper, a demodulation system for FBG micro-strain sensors based on back propagation neural network (BPNN) algorithm is proposed. The complex nonlinear relationship between the transmitted light intensity and the center wavelength of FBG sensor is established by using BPNN, and the absolute interrogation of the center wavelength of FBG sensor is realized. The demodulation performance is more advantageous compared with traditional machine learning algorithms, with an accuracy of ±9.3 pm. Through experiments, the effectiveness and superiority of the demodulation system are proved. It achieves the purpose of low cost, high performance and wide range of demodulation, and provides reliable analysis support for engineering applications.