Analysis And Research Of On-line Control Model For Copper Flash Smelting Process

Copper flash smelting is a complex continuous industrial process with nonlinear time varying, strong coupling and large time delay. The control accuracy has an important impact on the product quality and the life of the equipment of smelting. In view of the copper flash smelting process is mainly operated by manual mode and lack of accurate mathematics mode, the flash furnace changes frequently and is difficult to achieve stable production. On the basis of analysis of copper flash smelting technique and control process, blending optimization and online control mathematical model is studied. The main work and conclusions are summarized as follows.(1) Considering the concentrate composition fluctuation, a processing method for abnormal data of component detection value based on the hypothesis test of random variables is proposed, and the blending optimization of copper flash smelting process is studied, design and development an intelligent blending system based on weight. The system can integrate considering the major influencing factors such as the component, the smelting requirement and the storage of copper concentrates, have fast and accurate blending process and the calculation result can not only meet the requirement of smelting, but also consider the rational use of raw concentrates which could reduce the cost of production.(2) The metallurgical model based on material balance and heat balance is studied. Establish the material balance model, material compound presumption model, heat balance model. In the calculation of material balance, regression fit the relationship of Cu, Fe, S grade in matte and the relationship between Cu, Fe grade in slag and Cu grade in matte, assume the weight and composition of furnace ash and the impurity element distribution in matte and slag. Heat balance is calculated based on the compounds formation enthalpy.(3) The application of neural network model in the on line control of copper flash smelting is studied by using the improved back propagation algorithm. Network model which the input vectors only include the main elements and the input vectors also consider the impurity elements are established. Through training and testing the network shows that back propagation neural network model is applied to the copper flash smelting online control process has good generalization ability and accuracy. At the same time, the model training and testing error comparison between the two input model shows that consider impurity elements in input parameters is necessary in the model design.