The integrated energy system is an efficient way of utilizing energy in industry park. However, with the massive integration of renewable energy and disorganized charging of electric vehicles, the safe operation of this system faces several challenges. To address these issues, we propose a novel dispatch model that incorporates the flexible load characteristics of electric vehicles clusters. Firstly, we elucidate the operational framework for the integrated energy system in parks and establish models for users and microgrid operators incorporating carbon trading mechanisms. These models can effectively portray how an integrated energy system operates within a park setting. Secondly, using charging data from parks, we uncover potential dispatchable charging/discharging capacities for electric vehicles clusters and formulate strategies to utilize electric vehicles as flexible loads in our dispatch operation policy. By appropriately regulating electric vehicles charging/discharging behaviors, demand-supply balance within the system can be better achieved. Subsequently, aiming to maximize benefits for all entities in the park area, we construct a master-slave game model that involves multiple users and microgrid operators. Lastly, employing reinforcement learning concepts, we establish an equivalent power output models for wind turbines, photovoltaic power generation and apply it to an integrated energy system in an industrial park in a specific city. An analysis reveals that our proposed model not only minimizes cost associated with energy storage equipment but also significantly reduces carbon emissions; yielding mutual benefits for both microgrid operators and users.