Large-scale complex systems are characterized by a large number of interconnected variables and a diverse set of interactions. As the demand for the development and optimization of large-scale systems is growing, so does the need for better techniques to understand their underlying dynamic behavior and predict and manage their long-term performance. With the increased capabilities of computer technology, we have been able to run simulation models for these systems that are larger in scale and higher in complexity. While these advancements have enabled more accurate representations of real-world systems, the ever-increasing scale and complexity of simulation models may eventually result in models that are too complex to work with – giving rise to large-scale complex simulation models. In this dissertation, we aim to investigate to what extent the abstraction of large-scale complex simulation models, specifically discrete- event simulation models expressed in the DEVS formalism, can be automated using their state- trace data. In order to achieve this objective, we designed a method that integrates the fields of modeling and simulation and temporal data mining by utilizing state-trace data and applying frequent episode mining techniques to discover behavioral patterns.
|Award date||17 Jun 2022|
|Publication status||Published - 2022|