A modified RBM-based estimation of distribution algorithm for steelmaking-continuous casting scheduling

In order to solve steelmaking-continuous casting scheduling problem which is considered as a critical module of planning and scheduling in modern iron and steel enterprise, this paper propose a data-based evolutionary algorithm. Firstly, a mixed integer programming model considering practical constraints and multi-objectives is provided. Secondly, we develop a modified estimation of distribution algorithm based on restricted Boltzmann machine, which incorporates a backward list scheduling method, a solution representation with job permutation within each batch, and a local search method used to enhance the performance of this algorithm. To verify the proposed approach, a number of instances from real-world industrial data are generated. The final numerical results show that the proposed data-based algorithm is efficient and effective to solve practical steelmaking-continuous casting scheduling problems.