Infusing Autonomy in Power Distribution Networks using Smart Transformers

In this tutorial we describe a list of analytical methods from optimization, control theory, and machine learning that can be used for infusing autonomy in large-scale interconnected networks of microgrids with as little human intervention in their control loops as possible. The cornerstone medium for promoting this autonomy is smart transformers made out of solid-state technology, which serve the dual role of a power-electronic transformer with very fast switching capability, and, hence, with significantly smaller size than conventional magnetic transformers, as well as of a powerful computer or logic-machine that can make intelligent decisions via communication. Solid-state transformers (SSTs) are, in fact, being anticipated to be the backbone of tomorrow's modern distribution grid. Microgrids typically have three layers of controllers - namely, primary control (for switching and circuit-breaking decisions), secondary control (for voltage and frequency regulation, and synchronization), and tertiary control (where a centralized supervisory controller communicates with energy management centers for updating setpoints according to load demands, and other changes in the circuit). Operating these control layers in today's microgrids require professional knowledge of the system, and calls for the operators to manually configure each of the components. With technology advancing to low-cost microprocessor driven devices such as SSTs, we propose a suite of new control and learning approaches that exploit different `Internet of things' functionalities embedded inside the SSTs, and thereby guarantee at-scale resilience, reliability, fault-tolerance, and autonomy for tomorrow's distribution networks.