Multi-Scale Analysis and Control of Smart Energy Systems

点击翻译按钮获取中文摘要

One of the principal aspects of the SmartGrid is the introduction of intrinsically stochastic and distributed renewable energy production, which decreases the ability of semi-centralized controllers to effectively control the system. To compensate for this effect, distributed communication, monitoring, and control systems are introduced. The microgrid is the logical unit of operations and interaction of these components, as well as the interface to the utility grid. The complexity and multifaceted nature of the system even at the local scale makes the analysis, prediction and control of the microgrid's dynamics a challenging task. The project will conjugate power system control, design and simulation with advanced theoretical analysis and control to an innovative framework. The envisioned methodology will enable a number of novel applications in smart energy systems ranging from system adaptation to user behavior, user classification, and feedback to the consumer. The project will engage undergraduate and graduate students in the research effort, and will create new curriculum opportunities for UCI?s students.The complexity of the problem addressed in the project originated from the large number of interconnected heterogeneous sub-systems interact and contribute to the overall system behavior. This interaction manifests at different topological scales and abstraction levels. The sub-systems interact at the physical level, where the electrical signals travel through the physical interconnections. However, significant temporal and inter-component interdependencies exist at the logical state level, that is, a set of variables describing the current state of microgrid components and influential factors. Control of the logical level typically takes the form of scheduling and management frameworks. Most prior work considers either the physical or the logical domain, without providing a clear methodology to bridge these two interdependent domains. The project will use advanced Dynamic Programming, graph theory, and estimation theory to create a bidirectional flow of information and control between the physical and logical systems, where learning algorithms are designed to map physical and logical signals to higher level logical states.