As an important part of the new power system

the energy interaction and sharing of multi-microgrids are conducive to the consumption of renewable energy and the enhancement of multi-agents operation efficiency. Aiming at the multi-microgrids source-load uncertainty and data privacy problems

a multi-microgrids cooperative optimal scheduling method based on energy sharing is proposed. Firstly

a shared energy storage and microgrid economic dispatching model based on a multi-agent interaction framework is constructed. Then

the power pricing strategies for shared energy storage and the economic scheduling decisions for multi-microgrids are realized by deep reinforcement learning method and mathematical planning method

respectively. Finally

simulation data analysis shows that the proposed cooperative optimal scheduling method can quickly cope with the stochastic variation of source-load

as well as effectively reduce the operation cost of multi-microgrids.