The Researches On Optimization Models Of Flux Addition And Matte Addition In P-S Converter

Matte converting is an important part of smelting crude copper for matte smelting factory, which is the key factor for production and quality of crude copper. However, the blowing process of P-S converter is not in a completely sealed environment, when it has chemical reactions, the changing and transferring of heat, changes of material, fluid dynamics, etc. These characters severely restrict the achievement of actual detection operation. Nowadays, the process operations in matte converting of most of domestic copper smelting factories depend on the manual experience. Obviously, the operation optimizations in each aspect of matte converting not only set a good foundation for increasing the end-point judgment rate, but also exist important realistic significance to the development of our country's copper industry.On the base of understanding the current development status of copper smelting at home and abroad, the additions system of matte and flux are founded by understanding the production status and structure of P-S converter, calculation of the matte material balance and the matte blowing characteristic. Compare the value of flux prediction model with the value calculated by empirical formulas and the value calculated by material balance, we can see the virtue and weak point. Finally, database of matte converting is founded to provide and update the clean data for the data-driven model.In the software development process, researches for BP neural network are carried out on both MATLAB and VC++ platform. And then the adaptive BP network model is established on VC++ platform, which can be optimized by the algorithm itself. In order to improve BP neural network algorithm, data normalization and grey correlation analysis are used to make pretreatment of data. In the end, a model with data-driven forecasting and data storage is created, which is base on VC++ and SQL.