面向双碳目标的分布式电源与大电网协同优化配置的深度学习方法研究  

Research on Deep Learning Method for Collaborative Optimization Configuration of Distributed Power Supply and Large Power Grid for the Dual Carbon Goals

作　　者：王绪利 朱刘柱 施天成 顾悦 刘高维 Wang Xuli;Zhu Liuzhu;Shi Tiancheng;Gu Yue;Liu Gaowei(Economic and Technological Research Institute,State Grid Anhui Electric Power Co.,Ltd.,Hefei Anhui 230022,China;East China Electric Power Design Institute Co.,Ltd.,China Power Engineering Consulting Group,Shanghai 200063,China)

机构地区：[1]国网安徽省电力有限公司经济技术研究院,安徽合肥230022 [2]中国电力工程顾问集团华东电力设计院有限公司,上海200063

出　　处：《电气自动化》2025年第2期1-3,共3页Electrical Automation

基　　金：国网安徽省电力有限公司企业研究类项目“安徽电网中长期发展投入策略研究”(SGAHJY00GHQT2200035)。

摘　　要：面对日益严重的全球环境问题,中国提出了“2030年碳峰值和2060年碳中和”的双碳目标。分布式电源与大电网协同优化配置是实现双碳目标的基础。为此,提出了一种基于果蝇优化的卷积神经网络与双向长短期记忆融合模型。该模型学习非线性复杂关系,并利用之前的配置信息进行训练,以识别系统特征和协同配置之间的相关性。试验结果表明,所提模型的精度最高达到了99.72%,电力系统的环保运行成本也得到了有效控制。因此,将所提模型应用于双碳目标下的分布式电源与大电网协同优化配置是可行的。Faced with increasingly serious global environmental problems,China proposed the dual carbon goals of“Carbon Peaking by 2030 and Carbon Neutrality by 2060”.The coordinated optimization configuration of distributed power sources and large power grids was the foundation for achieving the dual carbon goals.Therefore,a convolutional neural network based on fruit fly optimization and a bidirectional long short-term memory fusion model was proposed.The model learned the nonlinear complex relationships and carried out trains using previous configuration information to identify the correlation between system features and collaborative configurations.The experimental results show that the proposed model achieves a maximum accuracy up to 99.72%,and the environmental operation costs of the power system are effectively controlled.Therefore,it is feasible to apply the proposed model to the coordinated optimization configuration of distributed power sources and large power grid under the dual carbon goals.

关 键 词：分布式电源 大电网 卷积神经网络 双向长短期记忆 果蝇优化 

分 类 号：TM734[电气工程—电力系统及自动化]