The microtremor survey method was used to detect the development position of unfavorable geological bodies such as karst and fissure channels under the railway road base. As to the problems in the inversion results, such as blurred interface between soil and rock mass, inaccurate location and boundary of anomalies, etc., the distributed fuzzy clustering algorithm was used to analyze the inversion data. The basic principles of the microtremor survey method and the distributed fuzzy clustering algorithm was systematically reviewed. Taking the karst survey of the existing railway subgrade in a certain section of the Anhui-Jiangxi Railway Ningguo Line as an example, the distributed fuzzy clustering of microtremor exploration was carried out for stratum stratification and karst cavern automatic division. The inversion data before and after distributed fuzzy clustering analysis were compared with the drilling results, it is found that the distributed fuzzy clustering algorithm can automatically and effectively delineate the interface and abnormal areas, and more accurately identify geological anomalies. It shows that this method can greatly improve the accuracy of microtremor inversion data, and provide reference for the design and construction of railway subgrade engineering.