Aggregator pricing and electric vehicles charging strategy based on a two-layer deep learning model

Abstract

In response to the increasing number of EVs(electric vehicles) being charged in a disorderly manner, this paper proposes a two-layer deep learning model based on aggregator pricing and EV user charging strategy. Firstly, the relationship linking the power grid, aggregator, and EVs is defined. Secondly, the pricing of the aggregator and the allocation of EV charging stations are divided into two stages. In the first stage, the aggregator sets the tariffs and sends them to the users, and the users receive the charging strategy. In the second stage, the aggregator adjusts the tariff on the basis of the charging strategy provided by the EVs and the actual situation of different charging stations and then releases it. Next, a two-layer deep learning model is built to solve the final pricing and EV charging strategy model, and the concept of humanity is introduced. The upper layer is the charging decision layer, which is solved by deep reinforcement learning. The lower layer is the charging station selection layer, which is solved by deep Q-learning. The method proposed in this paper can effectively improve the profits of the aggregator and reduce the charging cost of EVs.