Demand Responsive Dynamic Pricing Framework for Prosumer Dominated Microgrids using Multiagent Reinforcement Learning

TL;DR

This paper introduces a multiagent reinforcement learning framework for real-time demand response in prosumer-dominated microgrids, improving operational efficiency and economic outcomes over traditional methods.

Contribution

It presents a novel RL-based decision-making environment for dynamic pricing in microgrids, addressing uncertainties and customer disutility issues of traditional demand response techniques.

Findings

Enhanced economic benefits for grid operators and prosumers

Improved market balance through prosumer energy storage utilization

Outperforms traditional demand response methods in simulations

Abstract

Demand Response (DR) has a widely recognized potential for improving grid stability and reliability while reducing customers energy bills. However, the conventional DR techniques come with several shortcomings, such as inability to handle operational uncertainties and incurring customer disutility, impeding their wide spread adoption in real-world applications. This paper proposes a new multiagent Reinforcement Learning (RL) based decision-making environment for implementing a Real-Time Pricing (RTP) DR technique in a prosumer dominated microgrid. The proposed technique addresses several shortcomings common to traditional DR methods and provides significant economic benefits to the grid operator and prosumers. To show its better efficacy, the proposed DR method is compared to a baseline traditional operation scenario in a small-scale microgrid system. Finally, investigations on the use of prosumers energy storage capacity in this microgrid highlight the advantages of the proposed method in establishing a balanced market setup.