【摘要】 提出一种虚拟电厂多主体效益分配方法，从两个方面对传统Shapley值法进行改进。一方面，综合考虑各主体对联盟的资源投入、风险承担、减排贡献3方面因素，引入修正因子对传统Shapley值进行修正；另一方面，采用基于强化学习的抽样方法对主体效益进行近似计算，解决了参与主体数量较大产生的分配过程“组合爆炸”问题。并将所提模型应用于某虚拟电厂实例仿真，在保证计算准确性的同时，缩短了计算时间。更多还原

【Abstract】 This paper proposes a multi-agent benefit distribution method for virtual power plants, incorporating two major improvements to the traditional Shapley value approach. Firstly, a correction factor is introduced to adjust the classical Shapley value by comprehensively accounting for each agent’s resource input, risk exposure, and contribution to emission reduction within the coalition. Secondly, a reinforcement learning-based sampling method is employed to approximate benefit distribution, effectively addressing the “combinatorial explosion” problem caused by a large number of participating agents. The proposed model is applied to a case study of a virtual power plant, demonstrating enhanced computational efficiency without compromising accuracy.更多还原