Situation awareness (SA) of air traffic controllers is defined as their ability to perceive, comprehend, and anticipate the future state of elements in the airspace. Accurate and rapid prediction of SA is considered essential for ensuring the safe and efficient operation of the aviation system. The situation awareness global assessment technique (SAGAT) was applied to objectively measure controllers’ SA based on task characteristics. A predictive model combining particle swarm optimization (PSO) and XGBoost was developed based on six control performance indicators. A visual comparison was conducted between the proposed PSO-XGBoost algorithm and three widely used algorithms. The results demonstrate that the combined optimization approach achieved a prediction accuracy of 97.17%, effectively enabling precise estimation of air traffic controllers’ situation awareness. Furthermore, based on the external datasets of the training trajectory, the effectiveness and generalization of the model were validated, proving its transferability and robustness with real data.The findings confirm the feasibility of SA prediction and suggest that the proposed algorithm and indicators provide an effective approach for enhancing prediction accuracy.