Energy Management with Renewables and Electricity Market Forecasting

Speaker: Prof. Georgios B. Giannakis (University of Minnesota)

Time and Date: 10:00-11:30, Oct. 31, 2013

Place: Room 138, Physics Building, Handan Campus

Abstract

High penetration of renewable energy sources critically challenges the economic dispatch of current and future power systems. Supply and demand must be balanced at every bus of the grid, while respecting transmission line ratings and accounting for the stochastic nature of renewables. In par with this goal, this talk deals with a network-constrained economic dispatch approach and the required predictions of electricity prices. Robust energy management schemes are presented for microgrids featuring distributed generation and storage units. Stochastic programming tools are adapted to ensure that complexity scales well with the sample sizes involved. Uncertainty of renewable energy forecasts along with their transaction prices are accounted for. These prices are predicted using state-of-the-art machine leaning techniques applied to electricity market inference. Forecasting leverages the low-rank matrix of spatiotemporally correlated prices, captured through suitable kernels across pricing nodes and hours. The efficacy of the novel approaches is assessed using standard IEEE power grid benchmarks.

Details