Alloy Addition Prediction in Smelting Based on BP Neural Network

Based on the investigation of impact on products composition of alloy additions in converter smelting process, a neural network model is established for the influence in this paper. Using BP neural network theory, select the actual production data as training samples and the main elements of alloy additions as input data. There are 4 layers in this model, namely input layer, hidden layer and output layer. VC++ language is adopted to develop program which is tried on in production site, the prediction accuracy of established network model hits more than 95%. The result shows that the predition method has good accuracy and effectively solved the big problem of relationship between product composition and alloy addition, and can automatically choose the request of alloy addition for a given product ingredients.