数据中心分布式能源系统与储能增强型不间断电源的协同优化：一种基于模型预测控制的三层优化框架

Abstract The rapid expansion of data center workloads presents pressing challenges to energy sustainability. In data centers, distributed energy systems (DES) often face high operational costs and renewable output volatility, while the storage potential of uninterruptible power supply (UPS) systems remains underutilized. To address these challenges, this study proposes a three-level optimization framework that integrates energy storage-enhanced uninterruptible power supply (EUPS) with DES. The framework aims to minimize operational costs, optimize renewable energy utilization, and enhance system stability. At the first level, capacity planning minimizes annualized investment costs. The second level establishes an operational scheduling strategy that balances cost efficiency and renewable integration. The third level uses model predictive control (MPC) to regulate EUPS in real time, reducing state-of-charge (SOC) fluctuations. Case studies demonstrate that the proposed strategy reduces annual operating costs by 6.7%, decreases photovoltaic curtailment by 68.1%, and reduces SOC fluctuations from 23.62% to 18.04%. The coordinated mechanism presented in this study optimizes the use of data center storage resources, improving both the economic performance and operational reliability of the system, offering robust support for the green transition in data centers.