Abstract: In order to monitor ground deformation effectively in real-time, and find out irregular rules and trends of surface subsidence, thereby determine the sinking situation of the whole surface movement areas. This paper used MIMU and lays fiber grating sensors to build a distributed network for surface deformation data measurement. A new exponential type of wavelet algorithm combined time series analysis was proposed to do the wavelet threshold denoising processing for the data collected. The effect of denoising is obvious .Finally, kalman filtering was used to forecast deformation trend of the deformable body. The results indicate that the data obtained by this method are accurate and effective. This method can be applied to engineering practice of terrain deformation forecast.