Publications

Description

Authors: Mohammad Afkousi-Paqaleh, Mohammad Jafarian, Andrew Keane

Abstract: The volatility of market prices and the interdependence of multiple markets make it challenging for Distribution System Aggregators (DSAs) to model these prices. In this paper, a novel statistical model based on Gaussian process regression and mutual information screening technique is developed. This model is able to predict different market prices and quantify their uncertainty whilst incorporating the interdependence of different markets. The proposed model is employed to assist DSAs with market price modelling. Price scenarios for various markets generated by this model make it viable to formulate the optimal involvement of DSAs in multiple markets as a stochastic multi-step two-stage problem. Other than providing a set of scenarios that efficaciously model multiple electricity market prices, after the clearing of each market, the proposed model leverages market clearing results to improve the accuracy of price prediction of subsequent markets. Extensive simulation results on large price datasets demonstrate that the proposed methodology will result in a considerable increase in the profit of the DSA compared to state-of-the-art price prediction approaches.