天气数据驱动下基于深度主动学习的新型电力系统供需失衡风险快速评估方法  

A Fast Assessment Method for Supply-demand Imbalance Risk of New Power Systems Based on Deep Active Learning Driven by Weather Data

作　　者：王邦彦 皮俊波 王秀丽[1] 齐世雄 孙文多 黄启航 魏成骁 张小平[2] 徐新然 王志维 WANG Bangyan;PI Junbo;WANG Xiuli;QI Shixiong;SUN Wenduo;HUANG Qihang;WEI Chengxiao;ZHANG Xiaoping;XU Xinran;WANG Zhiwei(School of Electrical Engineering,Xi’an Jiaotong University,Xi’an 710049,Shaanxi Province,China;Department of Electronic,Electrical and Systems Engineering,University of Birmingham,Birmingham B152TT,UK;National Electric Power Dispatching and Control Center of State Grid Corporation of China,Xicheng District,Beijing 100031,China;State Grid Zhejiang Electric Power Dispatching and Control Center,Hangzhou 310007,Zhejiang Province,China)

机构地区：[1]西安交通大学电气工程学院,陕西省西安市710049 [2]伯明翰大学电子电气与系统工程系,英国伯明翰B152TT [3]国家电网有限公司国家电力调度控制中心,北京市西城区100031 [4]国网浙江电力调度控制中心,浙江省杭州市310007

出　　处：《电网技术》2024年第10期4050-4059,I0034,共11页Power System Technology

基　　金：国家电网有限公司总部科技项目(5108-202319067A-1-1-ZN)。

摘　　要：相较于极端气象,天气因素对电力系统供需平衡的冲击常被忽视,但多日无风无光等事件同样可带来保供问题。提出了一种天气数据赋能、深度主动学习赋智的新型电力系统供需失衡风险快速评估方法,较传统方法更高效、更准确。首先,考虑源网荷储多环节建立以天为尺度的电力系统生产模拟模型,以进行无风无光等异常气象下系统供需失衡的分析。同时,针对传统可靠性指标的不足,提出以天为尺度的新型分布式指标,并进一步利用风险曲线描述系统长期风险。然后,提出应用深度主动学习的电力系统风险快速评估框架,建立风险预测主网络和误差预测副网络耦合的双深度神经网络,并构建相应的损失函数和训练流程。最后,基于IEEE标准算例进行了效果验证与各方法对比,结果验证了其高效、准确、可拓展等特点。该研究为新型电力系统的风险快速评估提出了一种新颖有效的思路。Compared with extreme weather,the impact of weather factors on the supply and demand balance of the power system is often overlooked,but events such as no wind and no sunlight for many days can also cause problems in keeping the balance.This paper proposes a supply and demand imbalance risk fast assessment method for new power systems empowered by weather data and deep active learning,which is more efficient and accurate than traditional methods.First,a power system production simulation model on a daily scale is established considering the multiple links of source,grid,load,and storage to analyze the system supply and demand imbalance under abnormal weather conditions such as no wind and no sunlight.At the same time,because of the shortcomings of traditional reliability indices,a new daily-scale distributed index is proposed,and the risk curve is further adopted to describe the long-term risk of the system.Then,a power system risk fast assessment framework using deep active learning is proposed,a dual deep neural network coupled with a main risk prediction network and an error prediction sub-network is established,and the corresponding loss function and training process are constructed.Finally,the effect verification and comparison of various methods are conducted based on IEEE standard test cases,and the results verify its efficiency,accuracy,and scalability.This research proposes a novel and effective idea for fast risk assessment of new power systems.

关 键 词：新型电力系统 天气因素 电力电量平衡 风险评估 深度主动学习 

分 类 号：TM721[电气工程—电力系统及自动化]