Estimation of Regulation Reserve Requirement Based on Control Performance Standard

Abstract

This paper proposes a method to estimate the real-time regulation reserve requirement based on the NERC Control Performance Standard (CPS). This method is constructed via three steps: first, a Multiple Linear Regression (MLR) model is applied to abstract the relationship between CPS and regulation reserve and other system conditions using training observations generated from a load frequency control model; second, a stepwise method with cross validation is used to select the most relevant features of MLR; and third, the regulation reserve requirement is computed by the MLR model as a function of the predicted system conditions and target CPS score. The recursive least square (RLS) method is used to update the model parameters in an online environment. Testing on a single area automatic generation control model with load and wind data from CAISO 33% Renewable Portfolio Standard scenario indicates the method outperforms methods used in the industry today.