Abstract: By analyzing the subway monitoring data and establishing the corresponding prediction model, the possible deformation in the future can be prospectively predicted, so as to ensure the safe construction and operation of the subway. Taking the foundation pit project in Beijing as an example, the deformation monitoring scheme is introduced. Taking a monitoring point as an example, the wavelet analysis is used to de-noising the original monitoring data of a monitoring point. The time series analysis model and BP neural network model are respectively used to model and analyze the de-noised data, and the fitting value of the original data and the predicted value of the future deformation are obtained. The deformation results are obtained by using the coherent point timing InSAR Sentinel-1A satellite images at the same time. Finally, by analyzing the predicted value and the actual value of the two models, comparing with InSAR results, the advantages and disadvantages of the two models in the application of subway deformation monitoring data can be compared.

Key words: subway deformation monitoring, analysis of time series, neural network, time series InSAR