Slides
Video
Abstract
We formulate a risk-sensitive forward energy procurement problem with uncertain wind generation that penalizes costs and constraint violations via conditional value at risk (CVaR). The risk-sensitive formulation allows a system operator to model her tolerance to high uncertain costs and constraint violations. We further propose a primal-dual stochastic subgradient based algorithm to solve the energy procurement problem where the dispatch decisions update in each iteration based on samples of wind power output.