Dynamic performance analysis is essential for production systems facing random disturbances. In this paper, a vulnerability evaluation approach is proposed for smartphone assembly production systems with finite buffers under a resilient system analytic frame. Firstly, four important vulnerability indicators, namely Terminal Time Delay (TTD), Terminal Time Window (TTW), Bottleneck Time Delay (BTD), and Bottleneck Time Window (BTW), are defined to expound temporal and spatial attributes caused by disruptive events. Then, a recursive derivation approach of the queuing network model is presented to obtain a state-transition matrix, wherein machine reliability is also considered in the model. Afterward, the exact solutions of steady and transient vulnerability are evaluated based on state probabilities inference. Finally, numerical studies are carried out to validate the proposed method and translate it into a practical tool. An application program with vulnerability analysis and disturbance control functions is developed, embedded in the digital twin system independently developed by our team to solve practical problems.