In contrast to constructing the dynamical model in the traditional Euclidean space, the modeling based on moving pattern characterizes the dynamics of systems in pattern moving space. Firstly, a method of constructing multi-dimensional pattern moving space based on particle swarm optimization and quadtree partition space is proposed. For quantitatively characterizing the moving of multi-dimensional pattern, the interval vector is used to measure the multi-dimensional pattern class variable. Then, a multi-dimensional interval T-S fuzzy model is defined and used to construct the prediction model of the multi-dimensional moving pattern. Finally, the effectiveness of the proposed model is verified by the simulation based on the real sintering production process data.