The resilience of urban transportation system is an important basis for building a resilient city. The existing assessment methods are faced with the challenges of complex influencing factors, difficult quantification of resilience index and low accuracy of traditional single model. In order to solve these problems, this study proposes a dynamic resilience evaluation method of urban transportation system based on deep learning model. This method constructs a resilience assessment model of urban transportation system, quantifies and integrates dynamic influencing factors. Convolutional neural network (CNN) was used to extract the dynamic characteristics of the toughness index, combined with Bidirectional Gated Recurrent Unit (BiGRU) to explore the time correlation, and the attention mechanism was used to enhance the important features. Taking Zhengzhou metropolitan area as an example, the empirical analysis found that the overall resilience of the urban transportation system in this region was at a medium to high level and showed a slow rising trend (resilience index increased from 0.319 in 2013 to 0.347 in 2022, with an average annual growth rate of 8.611%). However, there are significant differences in resilience levels among cities within the region (Zhengzhou has the highest annual mean resilience index (0.802) and Pingdingshan has the lowest (0.105). Compared with the traditional model, the model shows better performance in terms of mean absolute percentage error (MAPE = 0.06275%), root mean square error (RMSE = 0.018698%) and coefficient of determination (R²= 0.9912), which verifies the validity and reliability of the model.