Abstract: To improve the timeliness and accuracy of risk assessment for locomotive safety, this paper focused on the operation behavior of locomotive crew members and designed and developed an LKJ data comprehensive analysis and management system based on cluster analysis. The system realized automatic collection of LKJ operation record data and locomotive microcomputer data via Socket communication technology, and integrated them into a manipulation behavior dataset, then adopted the fuzzy C-means clustering algorithm to classify the manipulation behavior datasets. It compared the classified subsets of manipulation behavior data with the standard datasets of manipulation behavior to obtain diagnostic results for specific operation behaviors of locomotive crew members, and proposed suggestions for multidimensional comprehensive analysis of diagnostic results, which identified the frequent and occasional problems that locomotive crew members encounter during train operation through comprehensive analysis, thus providing data support for enhancing the precision of on-site management.