Based on the four dimensionless factors derived by dimensional analysis, the dimensionless prediction model for fluid-to-particle surface heat transfer coefficients (h_fp) was fitted by multivariate regression analysis. After modified the model with the residual analysis method, the final corrected coefficient of determination values R2 reaches 0.973. Furthermore, the five groups of verification set data were used to verify the accuracy of the constructed model, the predicted value of the theoretical model had a high consistency with the calculated value of the measured data, and the average relative error was 2.79%. The research indicated that dimensional analysis combined with numerical simulation has advantages of good reliability and wide applicability on oil frying process h_fp predict, which can meet the needs of food processing and cuisine engineering.