Risk-limiting power grid control with an ARMA-based prediction model

This paper is concerned with the risk-limiting operation of electric power grids with stochastic uncertainties due to, for example, demand and integration of renewable generation. The main contribution is incorporating autoregressive-moving-average (ARMA) type prediction models for the underlying uncertainties into chance-constrained, finite-horizon optimal control. This uncertainty model leads to a more (compared to existing work in literature) careful treatment of correlation in time which is significant especially in renewable generation yet has attracted limited attention. The paper first discusses how the resulting chance-constrained optimization problems can be solved computationally and demonstrates the effects of the use of the proposed prediction models through simulation-based case studies with realistic data.