Abstract: Aircraft carrier flight deck aviation support is a critical process for generating and sustaining the combat capability of carrier-based aircraft. Its characteristics of high risk, strong constraints, and strong coupling impose stringent requirements on the safety, reliability, and interpretability of intelligent support systems. Focusing on decision support for aircraft carrier flight deck aviation support, this paper systematically reviews the research foundations in rule constraints, situational awareness, and scheduling optimization, and provides an in-depth analysis of the key challenges faced by large models in this domain, including trustworthy output, multimodal cross-domain fusion, multi-scenario task adaptation, and multi-process causal reasoning. Based on two representative scenarios, namely pre-launch collaborative support and contingency replanning during the recovery phase, the paper further summarizes key elements such as task objectives, input modalities, constraint conditions, output forms, and validation loops. On this basis, an overall technical roadmap for large models in high-safety-level scenarios is proposed, including trustworthy generation and external verification based on physical mechanisms and operational regulations, generalized multimodal unified representation and encoding, domain knowledge-driven cross-scenario adaptation, and causal reasoning with emergent collaboration for support process coordination, In addition, the requirements for lightweight deployment, local knowledge enhancement, and integration with existing command workflows under shipborne edge-computing conditions are further discussed. The application of large models to aircraft carrier flight deck aviation support hinges on reconciling the contradiction between probabilistic generation mechanisms and deterministic safety requirements, while strengthening physical constraint embedding, explicit knowledge modeling, and process-level causal reasoning. The technical framework proposed in this paper can provide theoretical support and methodological reference for the safe and controllable application of large models in complex military support systems.