Abstract: Although China’s abundant coal resources occupy a leading position in the world, the surface subsidence and ecological environment problems caused by mining are becoming more and more serious. Traditionally, the study of surface subsidence caused by mining mainly depends on the establishment of observation stations in the affected area, and the use of professional software to generate a variety of two-dimensional static settlement curves. However, with the advent of the Cesium framework based on WebGL technology and its defined 3DTiles specification, it provides an effective means for multi-dimensional visualization and application of coal mining subsidence. In this paper, the dynamic prediction model of subsidence deformation at any point on the surface is constructed by optimizing the piecewise Knothe time function and combining the probability integral method. The dynamic prediction API of working face subsidence corresponding to the dynamic prediction model is developed, which can provide the subsidence matrix to the client browser. According to the land subsidence matrix, the two 3DTiles data sets of Terrain terrain tiles and dense point cloud pnts tiles obtained by oblique photogrammetry are modified, and the multi-dimensional display of surface subsidence is realized in the multi-dimensional visualization platform of mining area based on Cesium framework. This study provides more scientific decision support for mining area planning and management, and opens up a path of technological innovation for dynamic prediction and multi-dimensional visualization of mining subsidence.