Abstract
In the development of automated systems in the field of industrial automation, engineers often have to work on many tasks under time pressure, including simulations. In order to support engineers in these complex simulation tasks, a framework is presented which coordinates and calculates simulations based on structured knowledge in a parallelized, decentralized way. Different approaches for a problem are simulated automatically and compared with each other. Therefore, complex problems, meaning large-scale ones that often consist of coupled problem parts, are decomposed into partial problems. Their approaches are subsequently processed separately each using the most suitable software tool. Cooperating software agents are utilized to achieve a flexible coordination of decentralized simulations. These autonomous software units communicate with each other to control the parallelized activities during a simulation project. The presented concept is evaluated based on the application fields of control design and multiphysics simulation. Hence, multi-agent systems (MAS) were generated prototypically using the developed framework. As an addition to the framework for parallel simulation, a concept for supplementing the MAS with a case-based reasoning component is presented. Using the case-based reasoning addition, the performance of the simulations was further improved. Due to the varying computation time of the different approaches, analyzed in parallel, results are often available earlier. The MAS is also capable of improving the quality of the results because of the automatic analysis and comparison of different approaches.
Users
Please
log in to take part in the discussion (add own reviews or comments).
