Abstract: This study investigates the rescheduling optimization problem in virtual cell (VC) manufacturing systems triggered by the dynamic arrival of new orders, a prevalent disturbance in make-to-order production environments. The arrival of new orders disrupts initial schedules and induces state transitions in parts or machines, challenging traditional heuristic rules that focus solely on static process or capacity constraints and fail to account for dynamic state transitions during rescheduling. To address this gap, the paper proposes a joint decision-making model that integrates process route reconstruction with state transition-guided rescheduling. The model optimizes three objectives: minimizing the maximum completion time, total queuing time, and total transportation time, with weighting parameters to balance their relative importance. A novel feature of the model is its introduction of dynamic constraints for schedulable states upon new order arrivals, enabling real-time adaptability. Additionally, the model employs an assignment problem framework to precisely compute queuing times, eliminating the need for assumptions about order arrival distributions, which enhances its practicality in complex manufacturing systems. To solve this high-dimensional and multi-constrained problem, a Differential Evolution Hybrid Algorithm Using State Transition Information (DEHAUSTI) is developed. This hybrid algorithm combines the global search capability of differential evolution (DE) with the local state analysis of finite state machines (FSM). DEHAUSTI incorporates a dual-layer encoding mechanism for part sequencing and machine allocation, state transition rules to evaluate rescheduling feasibility, and adaptive selection strategies, including greedy and simulated annealing approaches, to mitigate the risk of local optima. The experimental validation involves four large-scale random instances, comparing DEHAUSTI with four benchmark algorithms: genetic algorithm (GA), standard DE, fruit fly optimization (FFOA), and forest optimization (FOA). The tests are conducted at ten rescheduling time points to evaluate performance under varying conditions. Results demonstrate that DEHAUSTI significantly outperforms the benchmark algorithms. Specifically, it reduces the maximum completion time by 4.73% in initial scheduling and 4.43% in rescheduling, cuts queuing time proportions by 44.96% and 85.50%, and lowers transportation time proportions by 3.96% and 1.39%, while simultaneously increasing the proportion of effective production time. Further analysis reveals that initiating rescheduling earlier yields additional performance benefits, highlighting the algorithm's adaptability to dynamic environments. Case studies, including detailed Gantt chart analyses, illustrate DEHAUSTI's effectiveness in balancing machine workloads, reducing bottlenecks, and enhancing system flexibility. The research contributes both theoretically and practically by establishing a state transition-aware rescheduling framework and providing actionable insights for VC systems operating under dynamic disruptions. The methodology's robustness is confirmed through parameter sensitivity analyses and comparative benchmarks, underscoring its superiority in handling high-dimensional, constrained optimization scenarios. Future research could extend this framework to address other disturbances, such as machine failures or workforce changes, further broadening its applicability in industrial settings.

Key words: Virtual cell, Rescheduling optimization, State transfer, New order arrival, Queue time