| 引用本文: | 梁 晨,曾 博,雷乐意,等.算力-电力联合市场下数据中心与配电网集成规划：一种多目标区间-随机优化方法[J].电力系统保护与控制,2025,53(16):120-135.[点击复制] |
| LIANG Chen,ZENG Bo,LEI Yueyi,et al.Integrated planning of data centers and distribution networks under the computing-electricity joint market: a multi-objective interval-stochastic optimization approach[J].Power System Protection and Control,2025,53(16):120-135[点击复制] |
|
| 摘要: |
| 为解决数据中心在电力-算力联合市场与低碳配电网协同规划中的问题，提出了一种多目标区间-随机优化方法。通过引入算力租赁机制，优化了资源利用效率与经济效益，同时降低了运营成本与碳排放。针对协同规划中的多目标优化与高维不确定性问题，设计了一种基于分解的自适应约束处理区间多目标进化算法—采用两种交叉策略(adaptive constraint-handling interval multi-objective evolutionary algorithm based on decomposition with two crossover strategies, ACIMOEA/D-TCS)。该算法能够高效求解帕累托前沿，提供鲁棒性和可行性兼具的优化方案。结果表明，数据中心参与算力市场显著提高了资源利用效率和经济效益，同时有效降低了碳排放。通过对算力资源租赁与配电系统运行的优化，所提模型在经济和环境效益方面取得显著提升，为电力-算力联合市场下的协同规划问题提供了新的理论方法与解决方案。 |
| 关键词: 数据中心 算力市场 低碳配电网 多目标优化 区间-随机优化 可再生能源 |
| DOI：10.19783/j.cnki.pspc.241429 |
| 投稿时间：2024-10-25修订日期：2025-01-23 |
| 基金项目:国家自然科学基金项目资助(52177082)；北京市科技新星计划项目资助(20220484007) |
|
| Integrated planning of data centers and distribution networks under the computing-electricity joint market: a multi-objective interval-stochastic optimization approach |
| LIANG Chen,ZENG Bo,LEI Yueyi,WANG Han,WANG Yuan,ZHANG Jiayi |
| (State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources (North China Electric
Power University), Beijing 102206, China) |
| Abstract: |
| To address the challenges in the collaborative planning of data centers into the low-carbon distribution networks under the computing-electricity joint market, this paper proposes a multi-objective interval-stochastic optimization approach. By introducing a computing power leasing mechanism, the method optimizes resource utilization efficiency and economic benefits while reducing operational costs and carbon emissions. To tackle the multi-objective optimization and high-dimensional uncertainty issues in the collaborative planning process, an adaptive constraint- handling interval multi-objective evolutionary algorithm based on decomposition with two crossover strategies (ACIMOEA/D-TCS) is designed. This algorithm efficiently solves the Pareto frontier and provides robust and feasible optimization solutions. The results show that the participation of data centers in the computing power market significantly improves resource utilization efficiency and economic benefits, while also effectively reducing carbon emissions. Through the optimization of computing power resource leasing and distribution system operation, the proposed model achieves significant improvements in both economic and environmental benefits, providing new theoretical methods and solutions for collaborative planning in the computing-electricity joint market. |
| Key words: data center computing power market low-carbon distribution network multi-objective optimization interval-stochastic optimization renewable energy |
