Abstract: Visualization of astronomical particle data could greatly help the scientists to understand the structure and evolution of the universe. Temporal data visualization can show complex structural features in the process of simulation, which can't be achieved by other methods. In this paper, we analyze the existing particle rendering methods, and propose an optimal method based on the tetrahedral mesh. In the data processing stage of the optimal method, we divided the particles to N slices according to the particle ID, and at the same time, we calculate the movement state of the same particle at different times. On the stage of constructing tetrahedral, we use the particle data to construct tetrahedron and use the tetrahedron to construct mesh according to the periodic boundary condition and the particle's motion state. Then we employ the tetrahedral mesh to project and rendering, keeping the particles which construct tetrahedron unchanged at different times. This algorithm consumes less memory than the original one and it suits for parallel computing. We compare the visualization results generated by our algorithm with the particle projection algorithm. The new algorithm has a great advantage in showing the structure of the universe, and it can make the filament more coherent and complete.