Abstract: In response to the problems of low utilization rate of unstructured equipment fault overview data and difficulty in analyzing the causal relationship in railway maintenance of way equipment faults safety management, this paper studied a method for causal correlation analysis of railway maintenance of way equipment faults based on large model retrieval augment. It analyzed the characteristics of safety data for railway maintenance of way equipment, extracted the causal factors of equipment failure overview based on railway natural language large model, introduced large model retrieval augment technology, proposed a causal correlation analysis process for safety data of railway maintenance of way equipment, and excavated the correlation between causal factors and business data. A case study and comparative experiment were conducted on the safety big data platform of a certain railway group company. The results showed that this method can provide technical support for understanding the causes of railway maintenance of way equipment failures and reducing railway safety issues caused by equipment.