Multi-objective optimization of operating parameters based on neural network and genetic algorithm in the blast furnace

Abstract

Addressing the complexity and isolation of the blast furnace, field engineers generally operate the system according to their former experience. While stability and safety are the first priority, it is natural to see extra consumption of ores and fuels. Over the recent years, researchers have been searching for the optimal operation point within the blast furnace by mathematical methods that include mechanism and data-driven modeling. The reported models have seldom succeeded in offering proper operation parameters to optimize several goals at the same time, as the factors of the blast furnace are highly correlative, which cannot be analyzed separately. In order to get the conflicting targets such as the lowest production cost and coke ratio as well as the highest iron quality, a multi-objective optimization model for the blast furnace based on BP neural network and genetic algorithm is proposed. The model is basically constructed on two theories. One is the BP neural network which can provide a mapping rule for multiple inputs and multiple outputs, the other is the genetic algorithm whose core contents are NSGA-II algorithm and Pareto optimality that are fit for the multi-objective optimization goal. By optimizing the blast furnace system with this model, the plant is feasible to keep the iron making process efficient and stable simultaneously.