Academic Awards 2023 booklet

79 A guide to learning modules in a dynamic network The advancements in technology have made real-life systems increasingly complex. Everything we see around us is now a large-scale interconnected network of various dynamical sub-systems (modules), known as dynamic networks. This ranges from our brain, which is a complex interconnection of neurons interacting with each other by sending impulse signals, to high-tech machines that contain various optical and thermo-mechanical components working together to provide the highest quality functionality. Finding out mathematical models of the sub-systems in the dynamic network and their interaction pattern (from data) has been garnering grave attention over the current decade. These data-driven models are used for prediction, simulation, design, and monitoring of these interconnected systems. Despite the availability of a few modeling techniques developed over the past decade, these are applicable only for dynamic networks that are small-scale and satisfy impractical & restrictive conditions like lack of confounding variables. In this project, we make the step to introduce modeling approaches for large-scale dynamic networks under different possible practical scenarios. This was achieved by developing novel ideas that incorporate concepts across different domains like machine learning, graph theory, system identification, and signal processing. For the first time, by assembling all the introduced methods in my research, we provide a wholesome decision flow-chart (Figure 1) that serves as an easy guide to every experimenter who wishes to effectively perform data-driven modeling of large-scale interconnected systems. Also, to aid the users from academia and industry, effort has been invested in the development of a toolbox. Having developed the groundbreaking fundamental approaches that can be applied to any application under different practical scenarios, we have now opened the path for the industry to use them in a diversity of applications like fault diagnosis, fault isolation, designing controllers for large-scale interconnected systems, and the development of digital twins. Figure 1: A guide for learning mathematical models in a dynamic network. The number in red specifies the chapter number of the PhD thesis that contributes to the decision chart.