Abstract: While the proportion of renewable energy connected to the grid gradually increases, the user side load also increases year by year. As an effective way of distributed new energy consumption, microgrid also faces great challenges in its power balance. In order to give full play to the regulation potential of user-side flexible load, and further consider the benefit distribution problem when market players such as micro-grid operators, energy storage operators and adjustable users participate in electric energy trading, this paper proposes a multi-agent game optimization scheduling strategy for island micro-grid considering virtual energy storage system. First, according to the air conditioning load and the operating characteristics of electric vehicles, the heat storage characteristics of buildings and the charge-discharge characteristics of electric vehicles are analyzed, and a joint virtual energy storage system including air conditioning load and electric vehicles is established. Secondly, considering the different interest demands of the three parties of source, charge and storage under the market environment, the transaction mode and mathematical model of each subject are constructed with the micro-grid operator as the leader and the energy storage operator and user as the follower, and the distributed collaborative optimization model of one master and many slaves is established. Finally, a numerical example is given to verify the effectiveness of the proposed method, i.e., while taking into account the interests of multiple agents, it can effectively reduce the peak-valley difference and reduce the regional power system's dependence on traditional energy storage equipment, thus saving the investment cost.