Abstract: To determine the proportion (mass ratio) of similar materials for physical models in coal mining, orthogonal experiments with 3 factors and 5 levels were conducted using fine river sand, gypsum, lime and water as raw materials. A multiple linear regression model was established using SPSS software, and regression equations for the mechanical parameters of each test were fitted. The influence laws of sand-binder ratio, lime-gypsum ratio and water content on the density and uniaxial compressive strength (UCS) of similar materials were obtained, and the optimal proportion for similar material tests was determined. Results show that the sand-binder ratio plays a dominant role in controlling the density and UCS of similar materials. At a significance level of 5%, the sand-binder ratio has an extremely significant effect on density and UCS, whereas the lime-gypsum ratio and water content exert no statistically significant influence but show certain regular effects. Both density and UCS decrease with the increase of sand-binder ratio, and first increase slightly with the rise of lime-gypsum ratio before decreasing after reaching the maximum value at a ratio of 5∶5; density rises initially and then falls with increasing water content, peaking at a water content of 15%, while water content has no obvious regular effect on UCS. A multiple linear regression equation model was established based on the orthogonal experimental results of 3 factors and 5 levels, with the level values of each factor taken as independent variables. Standardized residual analysis indicates that residuals are approximately symmetrically distributed in the vertical direction, and the predicted residuals of the model are relatively uniform at each fitted Y-value level without significant trends or skewness. The corresponding proportions for No.13 coal seam and its roof and floor in Panbei Mine were determined: sandy mudstone adopts the 5th proportion (fine river sand∶calcium carbonate∶gypsum=10∶7∶3), medium-grained sandstone adopts the 7th proportion (fine river sand∶calcium carbonate∶gypsum=8∶7∶3), and fine-grained sandstone adopts the 18th proportion (fine river sand∶calcium carbonate∶gypsum=8∶4∶6). The derived multiple linear regression equations can be used to predict the density and UCS of similar materials under different proportion conditions.