In the demodulation process of Fiber Bragg Grating（FBG）sensing system

the accuracy of the system is affected by the noise seriously.In order to improve the measurement accuracy of the system

denoising is an indispensable part in the demodulation.In this paper

we construct a new threshold function with much better denoising performance based on typical wavelet threshold denoising algorithm with soft and hard threshold function

and threshold functions in the existing literature.Through the de-noising experiment for the noisy FBG reflected signal

the result shows that the proposed threshold function can increase the average signal-to-noise ratio by 3

5

dB

and reduce the root mean square error by 0.1

0

.3

when comparing with the typical soft and hard threshold function.Finally

we apply the proposed method to the FBG temperature sensing system to demonstrate the feasibility of the method.The experiment results show that the demodulation accuracy is less than5 pm

which meet the accuracy requirement of the FBG sensing system.