The process of copper converter smelting has very complex nonlinear relations, so that its mathematical model can not be created accurately. In this paper, considering that the LS-SVM has strong nonlinear approximation ability, the organization and optimization model based on LS-SVM to assist copper smelting production is proposed, namely compensating the original model and improving the precision of the model using LS-SVM. To simulate using real production data of a copper smelting company, the result shows that this model’s precision is high, this can be used to guide the practice production and it is very effective in civil engineering