In the large capacity FBG sensor network demodulation system,signal denoising is an important factor that affects the accurate demodulation of the FBG sensor system.In order to improve the accuracy of demodulation,an improved variational modal decomposition combined with an adjustable semi soft threshold method for processing optical fiber signals is proposed in this paper.Four constraints are added to VMD to make its adaptive selected K values and reduce the calculation amount of VMD.Then denoising is performed with the new threshold function.Finally,through the temperature test experiment,the linear correlation coefficient between the wavelength and the temperature is obtained,and the performance and application value are evaluated.The experimental results show that the method in this paper achieves good results in both continuity and signal reduction.Compared with the traditional soft and hard thresholds,the signal to noise ratio (SNR) has increased by 6.0162 dB and 13.003 dB on average,and the root mean square error (RMSE) has been reduced by 0.0442 and 0.1487 on average.The linear correlation coefficient between temperature and wavelength in the temperature test experiment is 0.9992,which has better application value.