Evolutionary digital twin model with an agent-based discrete-event simulation method

A digital twin model provides the ability to adjust candidate behavior based on feedback from its physical part. However, small interactions between different subsystems and using real-time data about physical workshops are the primary problems in digital twin models. The essence of the digital twin model is the combination of the physical simulation method and the data-driven simulation method, and agent-based discrete-event modeling theory is an advanced way to build a digital twin model. Thus, the theoretical framework of the Digital Twin Workshop model is improved from the underlying modeling logic based on this new theory. By combining reinforcement learning with the digital twin workshop model, an evolutionary digital twin workshop model is developed in this study. This model is then applied to a real-world case. A comparison is made between the Digital Twin Workshop model with reinforcement learning policy and a heuristic policy and a random policy. The simulation results verify the validity and performance of the proposed model.