Collaborative optimization strategy of source-grid-load-energy storage based on improved Nash-Q equilibrium transfer algorithm

In order to give full play to the scheduling potential of multi-type energy storage resources and realize the collaborative optimization scheduling of source-grid-load-energy storage，a multi-type energy storage scheduling strategy including battery energy storage，pumped storage and electric vehicles is proposed. With the goal of low-carbon economy，the collaborative optimization scheduling model of source-grid-load-energy storage considering multi-agent game is established. In order to ensure the overall interests of source side，grid side and load side while taking into account their own interests，based on Nash equilibrium theory and using the reinforcement transfer learning technology，an equilibrium transfer algorithm based on improved Nash-Q is proposed. K-means clustering is used to discretize the data，and a dual-structure experience pool is added to improve the sample utilization rate，thus effectively improving the generalization ability of the model. Based on the data of an actual regional power grid，the simulative results show that the proposed strategy can effectively reduce the economic cost and carbon treatment cost of the system，and improve the new energy consumption capacity. © 2023 Electric Power Automation Equipment Press. All rights reserved.