Optimal peer-to-peer energy trading model with short-term load forecasting for energy market.

Energy trading and demand are key components of the electricity market, with accurate load forecasting essential for predicting consumption and optimizing costs. This research aims to enhance peer-to-peer energy trading (P2PET) through accurate short-term load forecasting (STLF). It addresses the challenge of forecasting future electricity load to facilitate P2PET, using the supply-to-demand ratio (SDR) method with stochastic integrated STLF. Machine learning-based regression methods are evaluated for STLF. The electricity load data used for STLF was collected from residential buildings in Brunei Darussalam. Comparative analysis identifies among several methods, LightGBM as the best-performing method, with an RMSE of 0.114, R2\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$${R}^{2}$$\end{document} of 0.74, MAE of 0.081, and a training time of 2.34 s. Hyperparameter-tuned STLF is used for effective P2PET, with energy trading prices determined via the SDR approach. Implemented in Python and MATLAB, the model demonstrates a 23.55% reduction in total energy costs, validating the impact of STLF on P2PET efficiency. [ABSTRACT FROM AUTHOR]