New SWRL framework enhances dynamic assembly scheduling using reinforcement learning

Researchers have developed a new framework called SWRL (Sliding-Window-Based Reinforcement Learning) to tackle the complex problem of dynamic assembly flow shop scheduling. This approach uses a graph-based Markov decision process to model multi-product kitting delivery, which introduces challenges in real-time scheduling due to dynamic order arrivals and changing job assignments. SWRL incorporates a sliding-window mechanism to filter irrelevant operations, a spatiotemporal graph network to track bottleneck shifts, and a dynamic action module to adapt to evolving action spaces. Experiments conducted with data from a home appliance manufacturer show that SWRL significantly reduces tardiness compared to traditional methods and existing deep reinforcement learning techniques, demonstrating robust performance across various resource and order configurations. AI

IMPACT Introduces a novel reinforcement learning approach for optimizing complex manufacturing scheduling problems.

RANK_REASON Academic paper detailing a novel reinforcement learning framework for a specific scheduling problem. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• deep reinforcement learning
• home appliance manufacturer
• Markov decision process

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →