In order to ensure flight safety and enhance operational efficiency, a classification method for civil aviation pilots' psychological health characteristics was developed based on machine learning clustering algorithms. Firstly, 215 active airline pilots were selected as research subjects. Psychological health data across two dimensions—psychological traits and psychological states—were collected using a pilot psychological competency assessment system. Subsequently, principal component analysis was employed for dimensionality reduction, and the elbow method was utilized to determine the optimal number of clusters. The K-means algorithm was applied to classify the samples into four distinct psychological health categories, facilitating the analysis of psychological characteristic differences among groups. Finally, visualization techniques were implemented based on the clustering outcomes to construct individual and group psychological health profiles. Radar charts were adopted to illustrate the distribution patterns of psychological characteristics across different pilot categories. The psychological health structure of pilots has group heterogeneity and diverse expression patterns. The proposed clustering algorithm can transform pilots' psychological health data into observable group feature portraits, achieving systematic identification and visualization of pilot psychological states at the operational level.