Decision-making related to production scheduling and roll change in hot rolling plays a crucial role in improving product quality and production efficiency. Conventional production scheduling often relies on fixed rolling lengths or slab quantities to reduce quality risks caused by roll wear, which overlooks variations in roll wear and the distinct roll profile requirements of different steel grades. To address these issues, this paper establishes a joint optimization model for production scheduling and roll change with the objective of minimizing total production costs, which include delay costs, quality costs, and roll change costs, by incorporating stochastic modeling and reliability assessment of roll wear degradation. This is done under the premise that all slabs within the same rolling unit require the same roll shape. The model comprehensively considers the quality grade requirements of different products, delivery deadlines, and the variability in roll wear. To solve this model, a multi-rule constrained dynamic adjustment factor genetic algorithm (MRCDAF-GA) is proposed. Numerical comparison experiments and sensitivity analysis using data from 431 slabs in actual production in a steel plant verified the effectiveness of the model and method. The proposed strategy achieved roll reliability between 0.867 and 0.939, reducing production costs by 6.51% compared with current practice. Furthermore, the proposed algorithm outperformed comparison algorithms in terms of convergence speed and running time.
Hu et al. (Tue,) studied this question.