CAREER: Stochastic and Data Driven Approaches for Addressing Variabilities in Power Consumption and Generation of Smart Distribution System

West Virginia University Research Corporation, Morgantown WV

Investigators

Abstract

This project will develop new computer-based methods to model and predict electricity distribution systems, so as to do a better job of handling uncertainty, multiple time intervals, and the possibility of greater penetration of plug-in hybrid (PHEV) and electric cars. PHEV are not expect to cause major problems for the major transmission grids, but they add major uncertainties and need for more flexibility at the level of distribution companies. To address that challenge, the PI will combine new work at the distribution level together with new mathematical methods for coping with uncertainty and random disturbances. She will also expand her ongoing efforts on behalf of West Virginia's "Vision 2020" strategic plan. The technical objective here is to identify stochastic models and develop controls for improving short term and long term performance of smart distribution systems. One often hears system stress, congestion and curtailment with volatility in renewable generation and demand. Transitioning to a sustainable energy future often results in complex and fast aggregated generator and load dynamics, which may lead to severe impact on distribution systems. This transition has forced power system researchers to pay serious attention to their time varying characteristics in a stochastic sense. This research has been facilitated by the advances in stochastic control theory and recent boost in data mining applications. The goal of the project is to develop a unified framework of stochastic and data driven approaches by which these variabilities can be identified, not by traditional approaches of modeling and control but by generating scenarios available from historical data and optimal predictive control design. The main question is whether massive amount of time series data can be handled by the data driven approaches to capture volatility and variability of generation and demand. The PI will use knowledge and concepts from stochastic theory, control theory, data mining, computational intelligence for a non-conservative approach for smart distribution system optimization and control.