The thermal system of iron ore pellet grate-rotary kiln process characterizes complex control object, long and lagging adjustment process and mutual coupling of various thermal parameters, as a result, the process cannot effectively predict the changing trend of each thermal parameter during regulation and control process, while the adjustment does not meet expectations. In order to solve these problems, based on the production data of pelletizing process in an iron and steel enterprise, the methods of data mining and correlation analysis are used to deeply study the correlation between grate-rotary kiln hot air circulation system, and the relevant mathematical model is constructed. The results show that the significance between the burner parameters of the rotary kiln head (jet volume, combustion-supporting flow) and the temperature of the kiln head and kiln tail and the thermal parameters of the grate-rotary kiln hot air circulation system are all 0.000, the correlation coefficient is basically above 0.900, and the regression model constructed between the parameters has a small error, which has been verified by on-site production, among which the error of the regression equation K₅ = 858.534 + 0.027X₁ + 0.030X₂ + 0.027X₃ and K₆ = 337.735 + 0.047X₁ + 0.068X₂ + 0.124X₃ set up among the the temperature of the kiln head and kiln tail and blast furnace gas consumption X1, Coke oven gas consumption X2 and combustion-supporting flow X3 are only 4.9% and 2.4%, which can be used as a prediction model for field thermal parameters.