![]() A prediction, even not very accurate, can constitute a relevant contribution to applying these distributed control approaches because it is possible to approximately predict the system’s performance and the self-organised ecosystem’s behaviour, which is impossible to do presently. Due to the similarity of the values predicted and collected from the real system, the results are promising. The first one is that the utilisation of simulation-based DTs to predict the behaviour and, more specifically, a distributed manufacturing control system’s performance can be useful. It is possible to conclude two main ideas from the results of the proposed article. The first operation comes in handy before the insertion of a product, as it allows the client to check the next product’s type, whereas the second is used whenever a product is added to the production line, changes position, its processing ends or when the production plan is first created. On the other hand, through a POST request, whose message body must contain the system’s current state and the remaining production plan, one may ask the API to update the file. By sending a GET request with no specific message, the API is instructed to return the production plan, after reading the file. For that, the latter provides two services which endure concurrent access. It contains a file, that plays the role of a database, where the demonstrator’s current state and the remaining production plan are stored, and an Application Programming Interface (API), whose purpose is to interact with the database. The Integration Layer (IL) is responsible for making the physical system’s relevant information available to external applications. To do that, it must use the production line’s current status and the production plan in order to simulate the system to different time horizons. The authors believe that a simulation-based DT can be used to predict the system’s behaviour and performance. Hence, this research aims to study the utilisation of Digital Twins (DTs) to mitigate this aspect. One of the main obstacles is the difficulty to predict the self-organised behaviour of such systems and the impossibility to understand how the system will evolve. In the literature, some authors already identified barriers as critical in adopting these approaches. Although the benefits and improvements introduced by these solutions are known and understandable by the manufacturers, it is still difficult for these approaches to be adopted and used in real industrial environments. Usually, these distributed control systems are modelled as modular components and designed using Multi-Agent Systems (MASs) approaches. Most of the proposed CPPSs aim to bring dynamic and flexible solutions using adaptable and reconfigurable distributed control systems. Whenever applied for manufacturing purposes, CPSs are referred to as Cyber-Physical Production Systems (CPPSs). Thanks to the latter, several conclusions and guidelines for future implementations were possible to derive as well. ![]() Furthermore, the digital model’s calibration is described jointly with the careful validation process carried out. Taking the aforementioned into account, the authors suggest an architecture that enables the integration between a previously designed and developed agent-based distributed control system and its DT, whose implementation is also provided in detail. Given their early stage, however, the already existing implementations are far from standardised, meaning that each practical case has to be analysed on its own and solutions are often created from scratch. The Fourth Industrial Revolution is leading to the emergence of new concepts amongst which DT stand out. ![]() Hence, the proposed research aims to explore the utilisation of Digital Twins (DTs) to predict and understand the execution of these systems in runtime. One of the main challenges is the unpredictability of the proposed solutions and the uncertainty associated with these approaches. Although these new approaches aim to deliver more flexibility and adaptability to the shop floor, they are not being readily adopted and utilised by the manufacturers. ![]() During the last years, several research activities and studies have presented the possibility to perform manufacturing control using distributed approaches. ![]()
0 Comments
Leave a Reply. |