Holonic Multi-Agent System to Develop Digital Twins Ecosystems

Abstract

Digital Twin (DT) is an emergent technology driven by the advances of Industry 4.0, being considered a significant component of the digital transformation. This approach relies on the integration of physical assets and their corresponding digital representations, ensuring the bidirectional synchronization and enabling the real-time monitoring and decision-support through, e.g., diagnosis, prediction, or optimization, based on advanced data analysis and simulation. The most common approach to design and implement DTs follows a centralized structure, considering the adoption of a digital representation that encompasses the entire asset under analysis. However, this centralized DT structure can be very monolithic, rigid, and inflexible to face the structural and behavioral complexity of the asset, as well as to properly address the assets life cycle. Insights from the Multi-Agent Systems and holonic principles can be used as inspiration to develop distributed organizational structures of DTs, taking advantage of their inherent characteristics of autonomy, cooperation and composition to build networks of interconnected DTs. Therefore, this paper proposes a holonic agent-based distributed DT architecture to face these requirements, framed with the ISO 23247 standard, and presents the design principles, dynamic behavior of individual DTs, and interaction patterns to enable interoperability between individual DTs as part of an ecosystem. A case study related to a modular conveyor transfer system is used to illustrate the application and validation of the proposed approach.