Abstract: Abstract: Flax is one of the five major oil crops in China. However, the local planting areas are mostly mountainous and hilly landform, and the harvesting process is excessively dependent on manual operation. Due to the problems of high labor intensity and low rate of mechanized equipment for harvesting operation, the development process of flax industrialization in China is seriously restricted. At present, the flax harvest process is still dominated by segmented harvest, and the threshing material formed after flax harvest needs to be separated and cleaned to remove the impurities in the grain. According to different principles, grain separated process can be divided into vibration screening and pneumatic sorting. Discrete element modeling technology is an auxiliary research method widely used in vibration screening research, which can establish various material shapes in agricultural production. However, the existing research methods use the multi-spherical aggregation method to fit the shape of grain and stalks, which is different from the real material in structure. On this basis, in order to improve the shape reality of flax threshing material and reduce the simulation modeling error, based on the open source discrete element method framework, the non-spherical particles of different components of flax threshing material were established by using the super-quadric surface element method, and the multi-stage vibration screening process was simulated and experimentally studied. In this study, the distribution of materials on the screen surface of the first screen and the second screen, the percentage passing and impurity content of different materials were analyzed, and the experimental verification was carried out. Different shapes of flax threshing could be constructed by changing the parameters of the super-quadric control equation. According to the volume fraction of different components of flax threshing, 3 148 grains, 1 728 stems and 124 fruits were generated in the simulation, respectively. In order to facilitated the analysis of simulation data, the primary screen and the secondary screen were divided into three different data statistical areas. The vibrating screening process was mainly analyzed from the material distribution on the screen surface, the screening rate of different materials and the impurity content. The results showed that the non-spherical particles of super-quadric surface could well simulate the physical properties of real materials; The percentage passing of seed and short stalks in the first screen was 99.08% and 66.49%, respectively, and the whole fruit could be sorted; In the second screen, the percentage passing of seed and short stalks were 96.38% and 4.79%, respectively, and the short stalks could be sorted. The sorted material could migrate on the corresponding screen surface under the excitation, which was conducive to screening. The impurity contents of different materials were counted, including 4.87% for seed, 21.95% for short stalks and 3.38% for whole fruit. Then the simulation process was experimented and verified. The composition and quantity of flax thresher in the test were consistent with the simulation, and the different components were mixed evenly before the experiment. In addition, the equipment must be subject to idle commissioning before the threshing material were dumped into the first screen. After the experiment, counting the quantity of different materials from the bottom collecting tank to the first screen. The experiment results showed that the relative data deviations of percentage passing and impurity content of different components of flax threshing materials were less than 10%. Although the modeling method used in the study has good shape fitting degree for non-spherical particles, it still had some limitations, which were discussed in this study. The results showed that the non-spherical particles established by the super-quadric surface method could be applied to the simulation of non-spherical materials involved in different agricultural production process.