Intelligent scheduling for distributed-level island integrated energy systems considering multi-energy utilization and incentive-penalty stepped carbon trading mechanism

Due to geographical constraints, island regions at edge distribution networks generally face challenges of resource shortages and high carbon emissions. To enhance resource utilization efficiency, this paper proposes a multi-energy utilization module (MEUM) for distributed-level island integrated energy systems (IES). The module efficiently recovers and utilizes secondary resources generated during system operation, thereby providing additional economic benefits for the system. Furthermore, to incentivize system units to participate in carbon emission reduction, the incentive-penalty stepped carbon trading mechanism (IPSCTM) is introduced in the system operation stage, which enhances the willingness of units to engage in carbon trading and reduces carbon emissions. Meanwhile, the scheduling problem of island IES that simultaneously considers efficient resource utilization and carbon emission reduction involves numerous interrelated variables, where traditional optimization methods rely on accurate models or predictive information. Therefore, to avoid modeling and prediction, this paper proposes a model-free deep reinforcement learning (DRL) approach to deal with the island IES scheduling problem. To validate the effectiveness of the proposed island IES model and solution approach, simulations are conducted based on operational datas from a representative island in northern China. The simulation results demonstrate that the proposed model can significantly reduce both the total operational cost and carbon emissions. Moreover, the proposed solution approach outperforms other methods in terms of optimization effectiveness and computational time.