A Review of Machine Learning Models for Energy Trade Ranking and Optimization in Smart Microgrids

The penetration of more distributed energy resources, such as solar photovoltaics, wind generation, battery energy storage systems, and electric vehicles, has shifted conventional power distribution networks from centralised to smart microgrids. One of the key problems of such environments is the efficient trading energy between prosumers, i.e., that a surplus and deficit energy must be ranked, assigned and optimised in real-time considering economic, technical and operational barriers. Traditional methods of rule based and optimisation driven approaches may not be able to manage stochastic characteristics of renewable generation, dynamic pricing, aspect of demand uncertainty, and scalability requirements inherent in smart microgrids. In this context, machine learning has become a powerful tool to improve energy trade ranking and optimisation, through data driven decision making, adaptive learning and predictive intelligence. This review paper systematically explores the latest development of machine learning models for energy trading in smart microgrid, mainly focusing on trade ranking mechanism and optimisation strategies. The paper reviews supervised, unsupervised, reinforcement and deep learning approaches and their respective roles in the areas of price forecasting, prioritisation of participants in the process, demand-supply matching and improvement in overall system efficiency. Furthermore, the paper relates the scope of these models with respect to key objectives, including cost minimisation, profit maximisation, fairness to participants, grid stability and resilience under uncertainty. A comparative discussion is provided to analyse the model's performance, computational complexity, model scalability and practical deployment challenges. By synthesising your results across recent literature, this review paper identifies existing research gaps associated with real-time implementation, interpretability, privacy preservation and regulatory integration. The purpose of this paper is to act as an integrated reference for researchers and practitioners by detailing some of the current trends, limitations and future research direction for the application of machine learning in energy-trade ranking and optimisation in smart microgrids.