Fiber Bragg grating temperature calibration based on BP neural network

Abstract A novel method which applies BP neural network (BPNN) to temperature calibration of fiber Bragg grating (FBG) sensors is proposed and discussed. Processing and analysis of experimental data showed that this method fitted very well the complex relationship between the center wavelength of FBG and temperature which is approximately linear in room temperature whereas nonlinear in low temperature. The maximum absolute error and root mean squared error were respectively 0.9434 °C, 0.2102 °C in fitting and 0.8943 °C, 0.2081 °C in testing which verified the advantage of BPNN fitting compared with the previous polynomial fitting. The forecasting performance of BPNN was also satisfactory. The novel FBG temperature calibration method based on BPNN has considerable application prospect in FBG temperature measurement.