A Methodology for Architecting Digital Twins in Factory Automation

Abstract

The fourth industrial revolution (Industry 4.0) gathered substantial attention form academic communities during the past years. As a multi-disciplinary concept, Industry 4.0 aims at formalizing the integration of digital technologies like Internet of Things (IoT), Artificial Intelligence (AI), and Big Data with factory automation and robotics. This integration aims at improving the overall performance of the manufacturing systems. Even though the Industry 4.0 has arrived as a new concept, it includes topics and concepts that have been introduced previously like the Lean Management, among others. This doctoral research includes a systematic review to position technical trends and concepts that contribute to the foundation of Lean 4.0 concept. Besides other outcomes, this review shows that Digital Twins (DT) acts as a facilitator of the concept of Digital Factories (DF). In general terms, a DT consists of a physical system, a digital replica of the physical system, and a continuous exchange of information between these two systems as initially introduced by Micheal Grieves. Therefore, implementing DTs in factory automation will contribute to digitising the factories, which in return, contributes to implementing a Lean 4.0 concept.

Digital Twins technology can be applied in various domains including manufacturing, healthcare, and urban development. In manufacturing, the concept of a process digital twin requires structured coordination of various systems that operate concurrently and interact within the domain of digital factories. These systems may include shop floor systems, Manufacturing Execution Systems (MES), and Enterprise Resource Planning (ERP). For a DT to interact with these systems, and as demonstrated in this doctoral research, it is important to adopt a generic and flexible approach as these systems may change its interfaces or core algorithms.

Fundamentally, an essential element within the digital twins’ system is the information model. Traditionally, these systems relied on databases due to its technical performance. However, to meet their growing demands, a more robust, yet flexible technology is required. In this regard, knowledge-based systems may function as a catalyst for constructing process digital twins. Furthermore, ontologies can be employed as a backbone for these knowledge systems, as it has been demonstrated as a suitable technology during this doctoral research.

In terms of building DTs for manufacturing process, this research considers DT architecture as a comprehensive set of directives for constructing the system, identifying the components and subcomponents that constitutes the DT, and detailing the blueprint of the components’ interactions. In other words, the core objective of this research includes providing guidelines for building process digital twins that features qualitative attributes like modularity, scalability, reusability, interoperability, and composability. These attributes prove to be essential for satisfying the target of providing generic and flexible solution for digitising factories. In addition, and as a result of this research, the presented architecture includes implementation that constructs digital twins using a multi-view, multi-layer, and multi-perspective which asserted the importance of the aforementioned attributes.

Besides the benefits of simulation and monitoring manufacturing processes, an essential additional advantage of the manufacturing processes DT is its capability to serve as a synthetic data generator. As demonstrated in this doctoral research, this practise enables developers of AI systems to collect big amount of data without the disturbing the real process. In fact, this data holds great significance in today's industrial landscape where novel AI techniques can enhance the functionality of physical systems.

This thesis is a compendium dissertation, which includes five peer-reviewed articles.