Pipeline transportation of coarse-grained solid-liquid two-phase flow is suitable for long-distance transportation of coal, and currently there are more studies on pipeline abrasion, while there are fewer studies on the prediction of particle size refinement after particle abrasion and the optimization of rheological parameters in the optimization of the combination of coarse and fine particles in the pipeline transportation of coarse-grained coal slurry, and this article proposes to adopt the abrasion equilibrium equations for the refinement of coarse-grained coal slurry hydrodynamic conveying, based on the method of the synthesis of the experimental data and the theoretical analysis. The solution model of particle refinement is given, and the viscosity prediction formula when mixing coarse and fine coal samples is also given by theoretical derivation. The analysis of the experimental data shows that with the prolongation of the conveying time, the content of the coarser particles decreases significantly, the proportion of <0.074mm fine particles increases significantly, and the viscosity of the coal slurry has a tendency to increase, but the magnitude of the increase gradually slows down. Comparison of the predicted and measured values of the particle refinement model showed that when the slurry conveying time was 30 minutes and 50 minutes, the maximum deviation between the predicted and measured values of the model was not more than 10.00%, and the results of the data analysis showed that the crushing rate function was the power function of the average particle size of each particle level, and the viscosity increased with the increase of conveying time. For the optimization of particle size grading, the Herschel-Bulkley fluid model description is used to give a prediction model of viscosity after grading, and the existing experimental data are used to verify the model, with a coal slurry concentration of 58% and different proportions of coarse and fine coal samples blended, the minimum value of viscosity exists, and the viscosity prediction equation is given, and the results show that the most deviation between the predicted value and the measured value is 5.23%. The coarse particle content with the lowest viscosity can be solved by using the theoretical formula for the viscosity reduction effect of blending.