Multi-microgrid electric energy transaction pricing strategy and system based on reinforcement and imitation learning

Abstract

The invention discloses a main power grid pricing strategy and system based on a reinforcement and imitation learning algorithm, the pricing strategy comprises micro-grid electric energy demand prediction, self-generation of an expert knowledge imitation learning mechanism and a strategy learning mechanism, and according to the method, on the premise that each micro-grid utility function parameter of a main power grid is unknown, an optimal pricing stragety is formulated to maximize personal economic benefit. the optimal pricing strategy is learned by using a reinforcement and imitation learning algorithm, experiments prove that the optimal pricing strategy can be converged after certain iteration to reach the optimal state of the economic benefit, and compared with other strategy methods, the overall economic benefit of the market can be maximized, demand response is realized, and the pressure of electric energy supply and distribution is reduced.

Description

technical field [0001] The invention belongs to the technical field of power system data security and control, and relates to a pricing strategy for multi-microgrid power trading based on reinforcement and imitation learning. Background technique [0002] A smart grid is a typical Cyber-Physical System (CPS) that enables the two-way transmission of information and power among various entities, including power plants, end users, and each end user. Compared with the traditional grid, the smart grid ensures the safe, efficient and reliable distribution of energy. As a local power distribution system in smart grid, microgrid has attracted much attention in recent years because of its advantages of environmental friendliness and self-sustainability. To meet load demands, renewable generator sets are integrated into the microgrid. However, intermittent renewable generators cannot meet the load demand of the microgrid due to unpredictable environmental factors, especially during ...

AI Technical Summary

Problems solved by technology

[0005] However, there are still two main challenges when applying RL and IL (RIL) based methods in the electricity market: 1) The reward function in the microgrid electricity market environment becomes sparse due to the existence of the daily settlement mechanism; The power grid only obtains the economic benefits of the day at the time of daily settlement, which greatly reduces the learning efficiency of the RL method; 2) The performance of the RIL method depends on the quality of the expert knowledge dataset

However, in the field of smart grid electricity trading research, there are few well-recognized expert datasets, which makes it difficult for IL methods to be directly applied to the electricity market

Images