System Dynamics Meets Multi-Agent Systems: A Review of Hybrid Simulation Approaches

This paper presents the development of systems dynamics (SD), and agent-based model (ABM) integration. A co-occurrence map illustrates the progress from theory and policy-wise hybrid simulations to integrations of cyber-physical systems with performance intelligence. One of the earliest research was about modality alignment of SD and ABM followed by the development of frameworks to combine policy decision and evaluation support systems with an pool of behavior system and heterogeneous individual agents. New clusters define a new cue from theory to realistic implementation. With digital twins and ABM, the systems are synchronized with their physical counterparts in near real time. Within these systems, autonomous agents manage the adaptive control in an environment with an IoT system which provide them with data that streams in continuously that facilitate a shift from descriptive to prescribed control. The prominence of concepts from digital twin reflects the real time data and IoT of the Industry 4.0. With these concepts, the paper outlines a systematized methodology for modeling hybrid SD agents that is coherent, scalable and transparent. The proposed framework clarifies and promotes the development of explainable and trustworthy autonomous actors and provides an ease of application to various systems.