Robust-stable scheduling in dynamic flow shops based on deep reinforcement learning

Abstract This proof-of-concept study provides a novel method for robust-stable scheduling in dynamic flow shops based on deep reinforcement learning (DRL) implemented with OpenAI frameworks. In realistic manufacturing environments, events endanger baseline schedules, which can require cost intensive re-scheduling. Extensive research has been done methods generating proactive schedules to absorb uncertainties advance and balancing the competing metrics of robustness stability. Recent studies presented exact heuristics Monte Carlo experiments (MCE), both are very computationally intensive. Furthermore, approaches surrogate measures were proposed, do not explicitly consider metrics. Surprisingly, DRL yet scientifically investigated stage production planning. The contribution this article is proposal how be applied manipulate operation slack times by stretching or compressing plan durations. demonstrated using different shop instances uncertain processing times, stochastic machine failures repair times. Through computational study, we found that agents achieve about 98% result quality but only take 2% time compared traditional metaheuristics. promising advantage use real-time environments supports idea improving techniques.