Locational Marginal Price Forecasting: A Componential and Ensemble Approach

• Published in: IEEE transactions on smart grid Vol. 11; no. 5; pp. 4555 - 4564
• Main Authors: Zheng, Kedi, Wang, Yi, Liu, Kai, Chen, Qixin
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.09.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Biological system modeling; Data models; Economic forecasting; electricity market; Forecasting; Locational marginal price; model stacking; Numerical models; Predictive models; Robustness (mathematics); short-term forecasting; Stacking; tree-based regression
• ISSN: 1949-3053; 1949-3061
• DOI: 10.1109/TSG.2020.2985070

Short-term locational marginal price (LMP) forecasting is the traditional problem of market participants and other institutions maximizing their profit. Most electricity market organizers in the world release the data of LMP along with its three components, i.e., the energy, congestion, and loss components. The series of the three components have their own patterns and driving factors, and can be utilized to improve the accuracy of LMP forecasting. However, most existing studies have focused on direct LMP forecasting and have barely noticed this characteristic. In this paper, we aim to bridge the gap between the released data of the three components and LMP forecasting through a componential and ensemble approach. Three individual forecasting models are selected and trained for these components, and an ensemble framework that stacks the summation LMP results and the direct results is proposed to enhance the overall accuracy and robustness. Numerical experiments with real market data are conducted to show the good performance of this novel approach.