Exploring and grasping the temporal and spatial variation rules of passenger group's aggregation in the station has a positive effect on optimizing the scheduling of vehicles in the urban rail transit network, especially optimizing the organization and management of passengers under disaster conditions. In this paper, a density based spatial clustering of applications with noise(DBSCAN) clustering algorithm based on the Gaussian mixture model is proposed for the station passenger location data set with non uniform density distribution. Firstly, the Gaussian mixture model is used to process the density of data sets. Then, local clustering is performed on data sets with different density levels to determine the parameters of each density layer data set. The appropriate seeds are selected to expand the local cluster cluster. Finally, the clustering results of each density hierarchical data set are merged. Through the calculation of the standard and measured data, it is illustrated that the DBSCAN algorithm based on the Gaussian mixture model has better clustering effect for the passenger location distribution data with non-uniform density distribution.