Predicting alloying element yield in a ladle furnace using PCA-DNN model

The composition control of molten steel is one of the main functions in LF refining process. In this study, a feasible model was established to predict the alloying element yield using principal component analysis (PCA) and deep neural network (DNN). The PCA was used to eliminate collinearity and reduce the dimension of the input variables, and then the data processed by PCA were used to establish the DNN model. The prediction hit ratio for the Si element yield in the error ranges of [-1, 1], [-3, 3], and [-5, 5] were 54.0%, 93.8%, and 98.8%, respectively, while, those of Mn element yield in the error ranges of [-1, 1], [-2, 2], and [-3, 3] were 77.0%, 96.3%, and 99.5%, respectively, in the PCA-DNN model. The results demonstrated that the PCA-DNN model performed better than the known models, such as the reference heat method, the multiple linear regression (MLR), the modified back propagation (BP), and DNN model. Meanwhile, the accurate prediction of alloying element yield can greatly contribute to realize a “narrow window” control of composition in molten steel. The construction of prediction model for element yield can also provide a reference for the development of alloying control model in LF intelligent refining in the modern iron and steel industry.