Back close

New interactive agent based reinforcement learning approach towards smart generator bidding in electricity market with micro grid integration

Publication Type : Journal Article

Publisher : Elsevier

Source : Applied Soft Computing

Url :

Campus : Amritapuri

School : School of Engineering

Year : 2021

Abstract : In order to suit the needs of the dynamically changing electricity market, software developers have developed various tools taking in to account the need of artificial intelligence for the electricity market entities. Algorithms in artificial intelligence are often divided into either supervised, unsupervised and reinforcement learning approach. A reinforcement learning when compared to supervised and unsupervised learning makes use of agent to learn from interaction with an environment and receives rewards based on the action it takes. It either exploits or explores in finding a solution. In the deregulated power market, the GenCos are modeled as agent by which the GenCo learns the market environment as agent and explores to get profited The Multi-agent based simulation is an effective method to incorporate this sort of intelligence and for providing efficient communication among the market entities. Using Multi-agent system, the problem existing in electricity market can be reduced since each entity problem can be solved by an individual agent. Multi-agent based reinforcement learning algorithm is used to handle the electricity market data. Here an agent based computational framework named Agent Based Modeling of Electricity Systems (AMES) under Java platform is developed for the design of electricity market. Market Agents balances the supply and demand through Market Clearing Price (no congestion) and Locational Marginal Price (congestion management) by performing optimal power flow. The agent also maximizes the profit of the Generator Companies (GenCo’s) through new learning strategy proposed using Variant Roth–Erev (VRE) interactive reinforcement learning method towards smart bidding among GenCo’s. The congestion relieving action in the transmission line and its effects on GenCo learning is discussed in this paper. The analysis is carried out on the electricity whole sale market functioning on a day-ahead basis developed by means of location and timing of injection of power. IEEE-3 bus system and IEEE-30 bus system with microgrid considered as non-dispatchable load is considered using agent based analysis. This technique helps the GenCo’s to attain possible high net earnings even with microgrid integration thus helps to relieve the congestion in the transmission lines.

Cite this Research Publication : Kiran P., Vijaya Chandrakala K.R.M., New Interactive Agent Based Reinforcement Learning Approach towards Smart Generator Bidding in Electricity Market with Micro Grid Integration”, Applied Soft Computing Journal – Elsevier, Part A, December 2020,

Admissions Apply Now