Baseline load forecasting using a Bayesian approach

Independent system operators (ISOs) and utilities have begun to realize the benefits of relying on demand response (DR) for service procurement. However, an aggregating entity in the command and contracting architecture should be included to provide the ISO flexibility in scheduling units. Demand baseline establishment is a key aspect in DR programs. Real-time updating techniques that account for uncertainty are needed to address baseline forecasting. In this paper, a Bayesian model is proposed to predict the load baseline and is formulated in a recursive Bayesian framework. Sample simulations have been performed on two test cases to illustrate the effectiveness of the proposed method and the benefits of implementing an online learning algorithm for baseline analysis to improve aggregators' risk averse bidding strategy.