Information Utilization Strategies Integration-Enhanced Rule Decision Framework for Rule-Based Real-Time Scheduling in Make-to-Order Flexible Job Shops

In make-to-order (MTO) flexible job shops, frequent and dynamic job arrivals place stringent demands on the real-time responsiveness of scheduling systems. Dispatching rules are widely used in real-time scheduling because they can generate decisions within milliseconds, but their performance depends strongly on how information is utilized at decision time. Using more up-to-date job shop information improves information timeliness, whereas incorporating information on the predictable release times of subsequent operations of unfinished jobs improves information completeness. However, the conventional assign-then-sequence (A-T-S) decision framework fixes machine-assignment decisions too early, making it difficult to combine forward-looking reservation with real-time revision. This paper proposes an information-utilization-enhanced rule decision framework for MTO flexible job shops. A sequence-then-assign (S-T-A) decision framework is developed to reorganize the timing and order of rule decisions, allowing predictive information on subsequent operations to be incorporated into the real-time decision scope. Because such forward-looking reservation may create machine idleness, a gap-filling mechanism is introduced to insert other processable operations into available idle intervals. Moreover, since the expanded information scope enlarges the rule element set and search space, multiple crossover operators are embedded in a gene expression programming (GEP) based rule-mining procedure to improve rule quality and search efficiency. Simulation experiments under six shop configurations show that the proposed framework achieves the lowest best-found and mean makespan values in all scenarios. In terms of mean makespan, the proposed framework reduces the value by 2.09–5.87% relative to the timeliness-enhancement-only variant and by 6.35–9.82% relative to the completeness-enhancement-only variant.