Seismic PSA is a critical tool for evaluating the seismic resilience of nuclear power plants, with its accuracy directly impacting the design and operational safety of nuclear facilities. In seismic PSA, the quantification of minimal cut sets (MCS) resulting from SRC50 is a key technical component. Traditional methods such as the minimum cut upper bound (MCUB) algorithm or multi-order approximation algorithms provide quick approximations but can result in significant errors in high seismic intensity regions. To improve the accuracy of seismic PSA, this paper presents an optimized quantification algorithm based on Binary Decision Diagrams (BDD). As a non-approximate method, the BDD algorithm theoretically offers precise quantification of minimal cut sets. However, its computational complexity grows exponentially with the size of the model, limiting its practical application. To address this issue, two optimization strategies are proposed: 1. Computation Process Optimization: By integrating the probability information of basic events into the BDD data structure and performing real-time calculation of cut set probabilities during recursion, the computational workload is significantly reduced. 2. Path Truncation Optimization: A threshold is set to prematurely terminate the computation of extremely low probability cut sets, further enhancing computational efficiency. Experimental results demonstrate that the optimized BDD algorithm not only maintains high precision but also significantly improves computational speed, providing a more efficient and accurate analytical tool for seismic PSA.