Social learning in self-organizing manufacturing networks based on opinion dynamics

The accelerated technological advancements in cyber-physical systems and edge computing have led to the maturity of self-organizing manufacturing systems. Many studies have addressed the optimization problem throughout the resource configuration process. With the current rapid growth in the number of manufacturing resources and the complexity of the relations among resources, it is necessary to analyze the utilization and workload of resources from a large-scale network perspective, in addition to the traditional optimization metrics such as time, cost, and quality. In this paper, a social learning framework was proposed based on the opinion dynamics models. Therefore, manufacturing resources can proactively share their states, e.g., busyness level, and negotiate their respective prices for use accordingly. The dynamic pricing mechanism was designed for better workload balancing as well as production pace management. Numerical experiments showed that the proposed methods can be easily integrated with other resource configuration algorithms to further optimize workload balancing and pace management.