Kriging interpolation algorithm is the basis of 3D modeling of coal seam occurrence form in fully mechanized mining face based on traditional geostatistics. However, the variation function model selected and fitted by Kriging interpolation algorithm cannot reflect the actual geological characteristics and the variation trend of spatial data. In this paper, UPSO-Kriging interpolation method is thus proposed: PSO algorithm is optimized to solve the problems of slow convergence and easily falling into local solution, and the optimized algorithm UPSO is then introduced into Kriging interpolation to solve the variation parameters and fit the variation function model, thus realizing the height prediction of each layer in the coal seam structure of working face. In addition, with DEM model established on regular grid method, the three-dimensional visualization of the occurrence form of coal seam in fully mechanized mining face is realized by using Three.js, which provides scientific basis for transparent mining, intelligent mining and quality mining of coal enterprises.