This paper addresses the optimization design problem of the Multi-mission Launcher (MML) system launch sequence, with a focus on the challenges of application in dynamic battlefield environments. The study proposes a Sensitivity-Guided Multi-Stage Sequential Quadratic Programming (SM-SQP) algorithm to overcome the limitations of traditional algorithms in adaptability, flexibility, and real-time performance. The SM-SQP algorithm combines sensitivity knowledge guidance with a multi-stage planning structure to adapt to information degradation conditions. Simulation results demonstrate that, compared to existing methods, the algorithm exhibits significant advantages in key performance indicators such as average time and ammunition consumption for destroying unit targets, as well as overall target destruction rate. These advantages are particularly evident in environments with interference and information degradation. This research not only provides a new perspective for optimizing the launch sequence of MML systems but also offers insights for addressing similar complex system optimization problems.