The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
Multi-agent Reinforcement Learning for Coordinated Production Scheduling
Abstract
This work examines the application of multi-agent reinforcement learning (MARL) for production scheduling in a real-world two-stage chemical production system found in industry. Improvement and automation of these scheduling practices have the potential to create tens of millions of dollars in savings. Reinforcement learning (RL) is promising for its powerful inference capabilities and data-driven flexibility. MARL methods provide a natural way of modeling coordinated systems and provide advantages for scaling and enhancing single-agent RL. Numerous MARL approaches exist, each with its own motivations, benefits, and drawbacks. Three selected MARL methods from the existing literature are applied to achieve dynamic scheduling in the described production system. The selected methods are evaluated by the quality of their solution and ability to provide effective production scheduling decisions. Well-studied optimization techniques such as mixed integer linear programming (MILP) have been frequently applied to solve similar problems. To benchmark the results of applying these MARL methods, each is compared against the results of a MILP model of the corresponding two-stage production process. Implications of each method on scale are considered with the objective of introducing concepts from multiple approaches and selecting a single recommended method. In addition to direct application within industry, the results of these experiments will serve to expand the set of available knowledge on applied MARL cases and provide new perspectives on the effectiveness of such methods for industrial scheduling processes.
Subject
reinforcement learningmachine learning
mixed integer programming
production scheduling
optimization
Citation
Camp, Andrew David (2022). Multi-agent Reinforcement Learning for Coordinated Production Scheduling. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /198485.