Abstract: Objectives The triggers of special situations during carrier-based aircraft takeoff and landing, as the origin of accidents, determine the development direction of special events and influence the decision-making of ship-based commanders. Characterized by scarcity and unpredictability, these triggers impose limitations on existing generation methods. While large models, with their extensive prior knowledge and strong logical capabilities, can alleviate these issues to some extent, they face severe hallucination problems in highly specialized scenarios like this. Leveraging the superior fitting performance of small models in professional contexts, this study aims to mitigate the hallucination issues of large models in generating triggers for special situations during carrier-based aircraft takeoff and landing. Methods To address the above challenges, this paper proposes a Small-and-Large Model Collaborative Driving method for generating special situation triggers (SGCAD). First, SGCAD constructs a knowledge base for carrier-based aircraft takeoff and landing using professional literature and retrieval-augmented techniques, forming a dataset of normal takeoff and landing descriptions. Subsequently, using normal operation descriptions as templates and incorporating scenario-specific prompts, large models are employed to generate potential triggers, while small models are utilized to distinguish between reasonable and unreasonable ones. Finally, SGCAD fine-tunes the large models via direct preference optimization algorithms, iteratively improving the proportion of reasonable triggers through multiple iterations. Results Experimental results show that after multiple SGCAD iterations, the proportion of reasonable triggers in the generated results reaches 94%. Furthermore, expert evaluation confirms that the generated triggers cover diverse complex scenarios, employ standardized terminology, and adhere to physical laws. Conclusions The proposed method effectively mitigates the hallucination issues of large models in generating triggers for special situations during carrier-based aircraft takeoff and landing, significantly enhancing the rationality and authenticity of generated triggers. This study provides a robust foundation for analyzing special situations in carrier-based aircraft operations and related research.