In order to maximise the reduction of pig iron cost in an ironmaking process, and at the same time ensure the output and quality of the pig iron, a design and optimisation system for the charging ingredients and structure in an ironmaking system was established using metallurgical theory. The system includes six modules, namely, sinter metallurgical performance testing and analysis, sintering burdening design, sinter component and property prediction, blast furnace burdening design, blast furnace batching calculation and ironmaking system burden optimisation. Based on actual production, testing and material balance theory, the system integrated these modules on VB and MATLAB using a series of intelligent algorithms, such as the BP neural network, multiple objective linear programming, genetic algorithms and so on. As a result, the optimum burden composition and structure of the sinter and blast furnace that satisfied all the constraint conditions could be obtained. Standing as a pinnacle of the global ironmaking production, the system can design and optimise not only the sintering burden, but also the blast furnace burden. Compared with the traditional production testing and hand calculation in the ironmaking system, the project can greatly reduce the production risk and greatly increase the calculation accuracy. Industrial application shows that the system is especially beneficial to reduce the ironmaking cost and at the same time ensure the output and quality.

In this paper, an optimisation system for charging composition and structure in the sintering process was established in order to reduce the sinter cost in the ironmaking process. The system comprised four modules: sinter metallurgical performance testing and analysis, sintering burdening design and optimisation, sintering matching calculation, and sinter component and property prediction. The data for the first module came from actual production values of a steelworks and from testing in the laboratory. Based on material balance theory, the second module used a linear programming method to optimise sinter cost, quality and quantity. The third module was built to predict the sintering production data. The fourth module can be used to predict some composition and properties of the sinter based on a Back-Propagation neural network. The system integrated all of these modules using Visual Basic and MATLAB. As the result, the optimum charging composition and structure of sinter which satisfies all constraint conditions can be obtained. Compared with traditional production testing and hand calculation in the sintering process the system can reduce the sinter cost and greatly decrease the production risk. Industrial application proves that the system is very useful and efficient in reducing sinter cost while ensuring output quantity and quality.