Abstract: In order to solve the problems such as vacuum interlayer leakage, pressure rise of the vacuum interlayer and rapid increase of inner pressure caused by vibration during the operation of cryogenic pressure vessel, a sequential prediction algorithm of vacuum failure degree of the cryogenic pressure vessel was designed by combining swarm intelligence algorithm and neural network. The process of vacuum failure of the cryogenic pressure vessel is simplified by simulation, the variation of various parameters in the process of vacuum failure of the cryogenic pressure vessel is obtained, and the failure data set is formed. The data set is used to train the LSTM model to predict the vacuum failure degree of the cryogenic pressure vessel. The hyper-parameters of the LSTM model are optimized by the particle swarm optimization algorithm. Finally, the trained PSO-LSTM model is used to predict the vacuum failure degree of the cryogenic pressure vessel in time series, and the prediction effect of the time series prediction model is analyzed.