Rule Construction and Simulation-optimization for Reentrant Batch Process Scheduling with Multi-device Joint Action

During the reentrant multi-purpose batch process with multi-device joint action, blockage, preemption and deadlock occur frequently, which significantly complicates collaborative production and increases the cost of changing production types. The direct application of intelligent optimization algorithms faces substantial challenges at both the encoding and decoding stages and fails to provide detailed and actionable scheduling plans. Therefore, an intelligent optimization framework integrating heuristic rules and high-fidelity simulation is proposed with the objective of minimizing the maximum completion time (makespan). Specifically, a production system simulation operation framework is established based on production logic, incorporating three rules—push-pull combination, heavy phase blending, and re-entrant priority—to ensure efficient and continuous production. Furthermore, equipment allocation rules under various order combinations are explored using genetic programming to guarantee high-quality decoding. Last, a genetic algorithm embedded with hybrid initialization and full-reachability mutation strategies is designed to optimize batch allocation and production sequencing. Experimental results from 1000 sets of orders across 28 varieties demonstrate that the proposed optimization framework surpasses high-fidelity simulation optimization methods based on industrial simulation platforms in terms of stability, effectiveness, and computational efficiency.