Bi-objective Reinforcement Learning Optimization of Virtual Power Plant Scheduling Considering Carbon Capture

As an integrated energy network with multi-energy flow interconnection, virtual power plant (VPP) faces some difficulties, such as excessive actual carbon emissions, insufficient absorption capacity of new energy, and a single optimization objective. Therefore, a bi-objective optimization scheduling method for virtual power plants is proposed to improve the new energy consumption capacity of multi-energy flow virtual power plants and reduce carbon emissions. Based on the objective function of minimizing the operating cost and carbon emission, a bi-objective proximal policy optimization (BiPPO) algorithm is proposed, which automatically finds the optimal scheduling scheme of VPP systems. To validate the effectiveness of this method, it is compared with other candidate algorithms in a VPP. Through experimental comparison, the effectiveness of the proposed method is confirmed.