Genetic Algorithm Based on Duality Principle for Bilevel Programming Problem in Steel-making Production

Steel-making and continuous/ingot casting are the key processes of modern iron and steel enterprises. Bilevel programming problems (BLPPs) are the optimization problems with hierarchical structure. In steel-making production, the plan is not only decided by the steel-making scheduling, but also by the transportation equipment. This paper proposes a genetic algorithm to solve continuous and ingot casting scheduling problems. Based on the characteristics of the problems involved, a genetic algorithm is proposed for solving the bilevel programming problem in steel-making production. Furthermore, based on the simplex method, a new crossover operator is designed to improve the efficiency of the genetic algorithm. Finally, the convergence is analyzed. Using actual data the validity of the proposed algorithm is proved and the application results in the steel plant are analyzed.

The crane is the major transport equipment for steel in the production process of steelmaking and continuous casting (SCC); it is responsible for shipping steel from the converter to the refining or continuous casting process. When a crane schedule problem happens, it will increase the redundancy waiting time between processes, and affect the close convergence between the refining process and continuous casting process. Considering unexpected events including time disturbance, temperature compensation, abnormal casting and steel grade changing, an optimal scheduling plan is not effectively obtained only by manual scheduling under the technical constraints of SCC production. The system is successfully applied to one large steel plant, and realizes rapidly optimal dynamic adjusting.Figure optionsDownload as PowerPoint slide