Real granular materials tests are commonly conducted to investigate the relationship between cushion particle’s shape, particle size distribution, porosity and bursting and puncture of the geomembrane on the cushion. However, rules of statistical significance are difficult to be obtained due to the randomness of the sample and the uncontrollable conditions of the experiment. To solve this problem, a stochastic reconstruction model of granular cushion is established based on Voronoi diagram, in which the particle size distribution and porosity of the model are controlled by adjusting the parameters of the random process. Statistical analysis was conducted for the shape parameters of the particles in the stochastic reconstruction model and their significance test between the real particles was also performed. The results show that the particle shape of the stochastic reconstruction model is close to that of the real cushion particles. The proposed model can be used to further study the bursting possibility of geomembrane on particle cushion and the relationship between the shape of the cushion particles, the particle size distribution and the porosity.