In view of the complexity of deformation analysis and prediction for the concrete dam, a deformation analysis and prediction method of concrete dam based on SSA-LSTM-GF is proposed by using the idea of phase space reconstruction and fusion modeling. Firstly, the measured data sequence of deformation is decomposed into the trend component, the periodic component and the remaining component by singular spectrum analysis (SSA), and the remaining component is removed as noise; then, long short-term memory (LSTM) model and Gaussian fitting (GF) algorithm are used to analyze and predict the periodic component and the trend component respectively; finally, the two results are reconstructed to obtain the final prediction results. The example analysis shows that SSA can achieve good data decomposition and denoising effect; LSTM model has superior prediction performance for periodic components; GF algorithm can well realize the fitting and prediction of trend components and the mining and extraction of some information; the reconstruction effect of the results of LSTM model and GF algorithm is good. In conclusion, the deformation analysis and prediction method based on SSA-LSTM-GF has certain feasibility and research application value.