To address the systemic issues of emergency medical resource allocation under multi-hazard coupling, this study constructed a hybrid rescue model combining fixed medical facilities and mobile rescue stations. Mixed-integer programming (MIP) was used to achieve three-dimensional optimization of “resource allocation-facility location-casualty transport,” with the objective function being to minimize the total rescue cost (including casualty transport time and waiting penalty costs). The uncertainty in disaster evolution is characterized using a scenario-based random demand representation method. Given the NP-hard nature of the model, PSO and VNS algorithms are designed to enhance solution efficiency through dynamic inertia weight adjustment and multi-modal neighborhood structure. Experimental validation confirms the effectiveness of the model and algorithms, providing practical insights for emergency management.