Cooperative Learning for P2P Energy Trading via Inverse Optimization and Interval Analysis

TL;DR

This paper introduces a cooperative inverse optimization method for P2P energy trading, enabling agents to learn optimal cost parameters within desired price and amount intervals, improving market outcomes.

Contribution

It presents a novel inverse optimization framework that allows peers to cooperatively determine their objective function parameters based on energy price and amount intervals.

Findings

Peers can set their parameters within analytically computed intervals.

The method ensures better alignment with desired energy prices and amounts.

Case study validates the approach's effectiveness.

Abstract

Peer-to-peer (P2P) energy systems have recently emerged as a promising approach for integrating renewable and distributed energy resources into energy grids to reduce carbon emissions. However, market-clearing energy price and amounts, resulted from solving optimal P2P energy management problems, might not be satisfactory for peers/agents. This is because peers/agents in practice do not know how to set their cost function parameters when participating into P2P energy markets. To resolve such drawback, this paper proposes a novel approach, in which an inverse optimization problem is formulated for peers/agents to cooperatively learn to choose their objective function parameters, given their intervals of desired energy prices and amounts. The result is that peers/agents can set their objective function parameters in the intervals computed analytically from the lower and upper bounds of their energy price and amounts, if the ratio of their maximum total buying and selling energy amounts lies in a certain interval subject to be learned by them. A case study is then carried out, which validates the effectiveness of the proposed approach.