Optimized Adaptive Scheduling of a Manufacturing Process System with Multi-skill Workforce and Multiple Machine Types: An Ontology-based, Multi-agent Reinforcement Learning Approach

The impetus for an interconnected, efficient, and adaptive manufacturing system, as advocated by the Industry 4.0 revolution, together with the latest developments in information technology, advanced manufacturing has become a prominent research topic in recent years. One critical aspect of advanced manufacturing is how to incorporate real-time demand information with a manufacturer's resource information, including workforce data and machine capacity and condition information, among others, to optimally schedule manufacturing processes with multiple objectives. In general, optimized manufacturing scheduling is a non-deterministic polynomial-time hard problem. Due to the complexity, scheduling presents a number of challenges to find the best possible solutions. This research proposes an ontology-based framework to formally represent a synchronized, station-based flow shop with a multi-skill workforce and multiple types of machines. Based on the ontology, this research develops a multi-agent reinforcement learning approach for the optimal scheduling of a manufacturing system of multi-stage processes for multiple types of products with various machines and a multi-skilled workforce. By employing a learning algorithm, this approach enables real-time cooperation between the workforce and the machines, and adaptively updates production schedules according to dynamically changing real-time events.