An improved simulated bifurcation algorithm for hot rolling mill production scheduling
将热轧调度问题抽象为多旅行商问题，基于伊辛模型提出改进的模拟分岔算法CSB，在基准问题和实际案例中比现有方法收敛更快、换辊成本更低。
The scheduling of hot rolling in steel production is critical to the industry's efficient development. However, it still faces challenges such as low scheduling efficiency and high rolling costs. The hot rolling scheduling problem requires efficient allocation of limited resources and time while meeting production demands, making it a typical combinatorial optimisation challenge. In recent years, quantum (heuristic) computing has demonstrated stronger potential than traditional methods in solving such complex optimisation problems. This study abstracts the hot rolling scheduling problem in steel production as a Multi-Traveling Salesman Problem (MTSP), develops a steel hot rolling scheduling model based on the Ising model, and proposes an improved simulated bifurcation algorithm, Cooling scheduling Simulated Bifurcation (CSB), to solve the model. In the evolution process of spin (position) variables, the CSB algorithm integrates jump convergence checking based on a cooling scheduling strategy, adaptive step-size control, and a derivative-sign substitution strategy, all of which significantly accelerate convergence and improve solution accuracy. Experimental results on benchmark Ising problems and real hot rolling scheduling instances show that, compared with existing hot rolling scheduling approaches in the production research literature, the proposed CSB algorithm achieves faster convergence and lower roll-change penalties while maintaining industrial feasibility.