In this paper, stochastic optimization of economic dispatch (ED) and interruptible load management is investigated using short-term distributional forecast of wind farm generation. Specifically, using the statistical information of wind farm generation extracted from historical data, a Markov chain-based distributional forecast model for wind farm generation is developed in a rigorous optimization framework, in which the diurnal nonstationarity and the seasonality of wind generation are accounted for by constructing multiple finite-state Markov chains for each epoch of 3 h and for each individual month. Based on this distributional forecast model, the joint optimization of ED and interruptible load management is cast as a stochastic optimization problem. Additionally, a robust ED is formulated using an uncertainty set constructed based on the proposed distributional forecast, aiming to minimize the system cost for worst cases. The proposed stochastic ED is compared with four other ED schemes: 1) the robust ED; 2) deterministic ED using the persistence wind generation forecast model; 3) scenario-based stochastic ED; and 4) deterministic ED, in which perfect wind generation forecasts are used. Numerical studies, using the IEEE Reliability Test System-1996 and realistic wind measurement data from an actual wind farm, demonstrate the significant benefits obtained by leveraging the Markov chain-based distributional forecast and the interruptible load management.
ASJC Scopus subject areas
- Computer Science(all)