考虑数据中心共享储能与计算负荷时空迁移特性的虚拟电厂优化运行方法OA

With the rapid advancement of large-scale artificial intelligence models,data centers are increasingly tasked with managing immense computational workloads.However,the exponential rise in computing power demands has intensified energy consumption challenges within these facilities.This paper proposes an optimal operation method for a virtual power plant considering shared energy storage and spatiotemporal migration characteristics of computational load in data centers.Initially,power consumption models and spatiotemporal migration potential models are developed to fully harness the operational flexibility of data centers.Subsequently,a virtual power plant model incorporating shared energy storage is introduced,optimizing the allocation of energy storage resources to achieve resource sharing and balance.Furthermore,a bilevel collaborative optimization framework is established between the virtual power plant operator and the data center.In this framework,the operator,at the upper level,generates market-based electricity price signals and demand response directives to guide adjustments in the data center's energy usage patterns.Meanwhile,the data center,at the lower level,optimizes its electricity load and computational task allocation,feeding back the adjusted energy load and response capabilities to the operator.Finally,case studies and comparative experiments demonstrate that the proposed model significantly reduces operational costs,enhances computational load distribution,and improves renewable energy utilization,effectively validating the feasibility of the proposed method and model.