Agent-based Energy Market toward the Optimal Trading in a Power Network

Power trading market among ordinary houses equipped with solar power generators is considered.Trading is mediated by several brokers who represent the buying or selling by each house. Objective is to investigate whether the optimal power usage is able to be realized by the optimal trading strategy of each broker. The optimal power exchange is given by a linear programming (LP) model,which purely considers the power flow without any price. Each broker has a pricing strategy to maximize the reward, and the trading amount is assigned according to the relative price. Agent based simulation results show that reinforcement learning is effective to obtain the pricing strategy with the effective usage of the battery under the environment of unsteady and unbalanced amount of the power generation and consumption. The obtained trading results in a similar power flow with the optimal power exchange derived from the LP.