The chemical composition of sinter is one of the key factors that determine the quality of sinter and the direct motion of blast furnace smelting. In order to stabilize the chemical composition of sinter, the chemical composition of sintered ore and mixed ore is analyzed, and the chemical composition with high correlation between the two is selected as the input parameters, and the Bayesian algorithm is used to optimize the hyperparameters in the random forest algorithm to predict the chemical compositions [w(TFe), w(FeO), w(SiO₂), w(CaO), w(MgO), w(Al₂O₃） and the correlation between alkalinity (R)] of the sinter, and a prediction model of the chemical composition of sinter based on BO - RF is established. The prediction results show that the prediction accuracy of the chemical composition of the sinter reaches more than 90.0%, of which the prediction accuracy of sinter w(CaO) is 96.9%, and the prediction accuracy of sinter R is 96.5%. The relative error of chemical composition prediction of sinter is within 7.0%, of which the relative prediction error of sinter R is within 3.5%, and the relative prediction error of sinter w(TFe) is within 1.4%, and the overall prediction accuracy is high.