The service life of pressure vessel is determined by creep and fatigue when pressure vessel is in high temperature for a long time. However, the existing methods have some defects in high temperature creep fatigue evaluation of complex models under multi-transient conditions, such as large calculation amount and low efficiency. Therefore, an efficient safety assessment method for high temperature pressure vessels was established based on ASME Ⅲ Division 5 code by combining finite element analysis with deep learning method. The strength check and creep fatigue damage safety assessment of the cylinder under a high temperature pressure vessel under multi-transient conditions of 400 ~ 500 ℃ were taken as an example, and deep learning was used for damage prediction and evaluation. The results show that under the premise of meeting the ASME code strength check, the maximum creep damage and maximum fatigue damage of the cylinder under the high temperature pressure vessel are 0.292 and 0.003 38 respectively, both of which are in the creep fatigue damage envelope. The R² values for the prediction of creep damage and fatigue loss are 0.999 8 and 0.983 9, respectively. The predicted creep fatigue damage is still in the envelope.