Stochastic unit commitment considering Markov process of wind power forecast

Abstract

Unit commitment (UC) is a major problem in power system operation which determines the operation schedule of the generating units by minimizing system operation cost. Because of the uncertainty of wind power, the UC problem needs to solve as a multi-period stochastic optimization. In this stochastic problem, scenarios tree is generated and may be too large to be solved when time horizon is longer. This paper presents an approach based on Maximum Entropy principle to generate and reduce scenarios by transforming a stochastic process to a finite-state Markov chain process and finding transition probability matrix. This approach is applied to transform a wind power process modeled by ARMA(1,1) model with Stochastic Volatility. A simple stochastic unit commitment is solved in this article. Because of power system security, reserve constraints also considered.