| 引用本文: | 陈晓娇,李 力,张秀青,等.复杂环境下海岛微电网源荷协同分层调度策略[J].电力系统保护与控制,2026,54(08):1-12. |
| CHEN Xiaojiao,LI Li,ZHANG Xiuqing,et al.Hierarchical source-load coordinated scheduling strategy for island microgrids under complex environmental conditions[J].Power System Protection and Control,2026,54(08):1-12 |
|
| 摘要: |
| 针对海岛微电网应对气候多变性与可再生能源波动性的运行稳定性挑战，提出一种基于空调负荷 (air-conditioning load, ACL) 调节特性的源荷协同分层调度策略。首先，融合轻量级梯度提升机 (light gradient boosting machine, LGBM) 与深度神经网络 (deep neural network, DNN)，构建 ACL 集群调节能力预测模型。接着，建立计及 ACL 集群调节补偿、弃光惩罚及复杂海洋环境随机扰动的双层优化模型。其中上层以日运行成本最小为目标，兼顾降低弃光率与激励用户参与；下层引入分布式加速控制算法以实现集群内部本地协同，在保障用户舒适度与隐私安全的同时提升收敛速度。仿真结果表明，所提方法日运行成本降低超过 12%，年运行成本降低约 13%，弃光率由 8.66% 降至 3.35%，并有效减缓储能退化，为海岛微电网在复杂气候条件下的经济高效运行提供支撑。 |
| 关键词: 海岛微电网 深度神经网络 空调负荷 环境因素 源荷协同 |
| DOI：10.19783/j.cnki.pspc.251129 |
| 分类号: |
| 基金项目:国家自然科学基金项目（52207034） |
|
| Hierarchical source-load coordinated scheduling strategy for island microgrids under complex environmental conditions |
|
CHEN Xiaojiao，LI Li，ZHANG Xiuqing，HUANG Liansheng，HE Shiying
|
|
1. Institute of Plasma Physics, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei 230031, China；2. School of Automation and Electronic Information, Xiangtan University, Xiangtan 411105, China
|
| Abstract: |
| To address the operational stability challenges in island microgrids under highly variable climate conditions and renewable energy fluctuations, a source-load coordinated hierarchical scheduling strategy based on the regulation characteristics of air-conditioning loads (ACLs) is proposed. First, the strategy integrates a light gradient boosting machine (LGBM) with a deep neural network (DNN) to construct a regulation capability prediction model for ACL clusters. A bi-level optimization framework is then established, considering ACL regulation compensation, photovoltaic curtailment penalties, and stochastic disturbances arising from complex marine environments. The upper level minimizes daily operating costs while simultaneously reducing curtailment and incentivizing user participation. The lower level employs a distributed accelerated control algorithm to achieve local coordination within the cluster, improving convergence speed while ensuring user comfort and data privacy. Simulation results show that the proposed method reduces daily and annual operating costs by over 12% and approximately 13%, respectively, decreases the curtailment rate from 8.66% to 3.35%, and effectively mitigates energy storage system degradation. These findings provide strong support for the economical and efficient operation of island microgrids under complex climatic conditions. |
| Key words: island microgrids deep neural network air-conditioning loads environmental factors source-load coordination |
