Multi-objective Evolutionary Algorithm for Production Planning in Cold-Rolling

The optimization of cold-rolling production plans is critical in practical steel production management, as it not only improves production efficiency but also ensures the uninterrupted operation of downstream production lines. However, this problem is highly complex due to the existence of multiple parallel cold-rolling production lines and the intricate many-to-many production relationships between these lines and their downstream counterparts. To further complicate matters, objectives are often conflicting, with manufacturers striving to increase production volume while simultaneously reducing cost. To address this challenge, we propose a multi-objective approach that minimizes both setup and outsourcing cost, which in turn contributes to both higher production volume and overall cost reduction. This approach combines a mixed 0-1 integer optimization model with a greedy-based multi-objective evolutionary algorithm (GMOEA) incorporating problem-specific evolutionary strategies. Experimental results, based on real-world data of different scales, demonstrate that these strategies significantly enhance the performance of the algorithm. GMOEA also outperforms other state-of-the-art multi-objective evolutionary algorithms that are widely used in real-world applications, confirming the effectiveness of the proposed approach, and demonstrating its potential for practical application in the steel industry.