Abstract: Equipment failures in aluminum coil production disrupt normal manufacturing operations, and how to adjust scheduling schemes dynamically in response to such failures has become a critical practical issue. To address this problem, this paper establishes a corresponding mathematical model and proposes a Kepler Optimization Algorithm with Adaptive Adjustment Based on Population State (KOABPS). Specifically, a hierarchy-based dynamic decoding method is constructed, Latin hypercube sampling is adopted for population initialization to ensure sample diversity, a population state-based adaptive adjustment mechanism is designed, and Levy flight is introduced to help the algorithm escape local optima effectively. To verify the effectiveness of KOABPS, a factory dataset is built using actual production data from enterprises, and random datasets of different scales are further constructed according to the characteristics of practical production data. Comparisons between partial rescheduling and complete rescheduling are conducted on the factory dataset, and ablation experiments are carried out to demonstrate the validity of the proposed key modules in KOABPS. In addition, small, medium and large-scale random experiments are designed to compare KOABPS with other typical optimization algorithms. The results show that the proposed algorithm outperforms its counterparts significantly in both solution quality and convergence speed.