Research on Temperature Calibration of Fiber Bragg Grating Temperature Sensor Based on BPNN Algorithm

Temperature calibration of fiber bragg grating (FBG) temperature sensor is of vital importance to measuring accuracy, especially in varying temperature conditions. To improve the measurement accuracy of the FBG temperature sensor, a new algorithm which adopts BP neural network (BPNN) to temperature calibration of FBG temperature sensor is proposed and analyzed. In this study, we illustrate the principle of FBG temperature sensing. To verify the performance of the proposed temperature calibration method based on BPNN algorithm, a series of calibration experiments under -55°C, -30°C, -5°C, 20°C, 45°C and 70°C were implemented. Comparison fitting results of polynomial fitting method and BPNN algorithm verified the BPNN algorithm could fit better the relationship between the wavelength of FBG and ambient temperature. The sum of squares due to error, mean squared error and root mean squared error are 0.0036 nm2, 5.9887 e-04 nm and 0.0244 nm in fitting result of FBG _ Tl sensor and 0.0024 nm2, 4.0096e-04 nm and 0.0199 nm in fitting result of FBG _ T2 sensor which proved the significant effectiveness of BPNN algorithm.