A Convergence Predictor Model for Consensus-based Decentralised Energy Markets

TL;DR

This paper presents a Bayesian logistic regression-based convergence prediction model for decentralized energy markets, capable of detecting potential cyber-attacks affecting market convergence with high accuracy.

Contribution

It introduces a novel Bayesian logistic regression approach for real-time convergence prediction in decentralized energy markets, enhancing security and reliability.

Findings

False rate below 0.01% on stressed datasets

Effective detection of cyber-attacks impacting convergence

Low-dimensional model operates efficiently for all prosumers

Abstract

This letter introduces a convergence prediction model (CPM) for decentralized market clearing mechanisms. The CPM serves as a tool to detect potential cyber-attacks that affect the convergence of the consensus mechanism during ongoing market clearing operations. In this study, we propose a successively elongating Bayesian logistic regression approach to model the probability of convergence of real-time market mechanisms. The CPM utilizes net-power balance among all the prosumers/market participants as a feature for convergence prediction, enabling a low-dimensional model to operate efficiently for all the prosumers concurrently. The results highlight that the proposed CPM has achieved a net false rate of less than 0.01% for a stressed dataset.