主动配电网态势图优化建模及大数据态势图片挖掘理论与方法研究

Active Distribution Network (ADS) has become a large dynamic complex network system with features of big data and high uncertainty, the operation of ADS is facing a great difficulty and risk consequently, therefore, an upper layer of situation management and decision making must be implemented upon the conventional real-time loop-closed control system, in which situation awareness visualization (SAV) is a core precondition because human being must be involved in situation management and decision making. However, there is a great lack of the basic application theory and methodology to make the SAV of ADS become practical. The first part of research on “Theory and Methodology of Situation Picture Optimization Modeling”, including the study of an unfiled modeling of information model and graphic model, the study of measurable assessment system of situation awareness visualization, and the study of a hierarchical optimization methodology, etc., is aiming at working out a “data-drive” automatic generation of situation visualization display for ADS with features of high aggregation of information (HAI), high identification of connection (HIC), and high perception of situation (HPS). The huge storage of the stored situation pictures, is also a highly integrated, refined and condensed big data of raw data, none exploitation and utilization of such a rich resource is a great waste. The second part research on “Theory and Methodology of Situation Picture Big Data Mining”, including the discovery and description of characteristics vector of situation pictures, the big data situation pictures clustering and analysis, and the big data situation pictures association relationship discovery and analysis, etc., is aiming at mining the situation knowledge or “re-perception” of the big data of situation pictures, and forming a situation perception mutual-learning mechanism between human being intelligence and machine intelligence, to help the refining of the design of situation picture, which can make sure of sustainable development of situation awareness visualization of ADS. To test the efficiency and practicability of the achievements from this research, a big data platform will be developed. All efforts are made to provide a solid theoretic and methodological support of industry application of situation awareness for ADS, accompanying some novel developments for the theory and methodology of big data technology itself.

主动配电网（ADS）为具大数据和强不确定性特征的大规模动态复杂网络系统，面临着巨大的运行难度和风险，须在闭环实时控制系统之上部署人类必须参与的态势管控层，因而态势可视化成为前提。ADS态势图优化建模理论与方法的研究，包括模数图一体化信息模型，态势图可测量评价体系，递级优化模型及算法等，以实现“数据驱动”的信息高集聚、接线高识别、态势高察觉态势画面自动生成。巨量存储的态势图片，形成生数据高度集成、精炼及浓缩后的大数据，值得挖掘；态势图片大数据挖掘理论及方法的研究，包括态势图特征发现与描述，大数据态势图片聚类方法与分析，大数据态势图片关联关系挖掘方法与分析，旨在形成人类智能与机器智能“态势察觉能力”交互学习机制，并可指导态势图改进，确保态势可视化的可持续性。搭建大数据平台，验证提出的理论与方法的有效性和实用性，从而为大规模ADS态势可视化提供应用理论与方法支撑，也为深化大数据应用提供新方法。

主动配电网（ADS）为具大数据和强不确定性特征的大规模动态复杂网络系统，面临着巨大的运行难度和风险，态势可视化成为前提。ADS态势图优化建模理论与方法的研究，包括模数图一体化信息模型，态势图可测量评价体系，递级优化模型及算法等，以实现“数据驱动”的信息高集聚、接线高识别、态势高察觉态势画面自动生成。巨量存储的态势图片，形成生数据高度集成、精炼及浓缩后的大数据，值得挖掘；态势图片大数据挖掘理论及方法的研究，包括态势图特征发现与描述，大数据态势图片聚类方法与分析，大数据态势图片关联关系挖掘方法与分析，旨在形成人类智能与机器智能“态势察觉能力”交互学习机制，并可指导态势图改进，确保态势可视化的可持续性。经过四年的艰苦努力，本NSFC基金项目发表了SCI/EI收录论文共29篇、科技核心杂志论文4篇；取得了6项发明专利授权和12项发明专利申请公开；取得了3项软件著作权；培养了19名硕士生；成果在1项国网总部项目、7项国网浙江省电力公司项目中得到了深入应用；为实际应用项目提供应用理论支撑，本NSFC项目凝聚了四个重点主攻聚焦方向的成果：（1）中压SCDN与态势图优化建模及大数据态势图片挖掘理论与方法，完整和系统地形成了解决一般县域中压配电网尤其是中大城市中压配电网的单线图自动生成理论与方法；面对中大城市的城区配电网，提出S-SCDN概念，并在单线图自动生成上取得重大突破。（2）低压TCDN与态势图优化建模及大数据态势图片挖掘理论与方法；在基于智能电表读数的低压配电网相线识别、表箱识别、电气拓扑识别等取得了重大突破；提出了面向不对称察觉的低压配电网分相接线图自动成图；（3）面向态势预测的配电网广义负荷预测，典型的如取得突破的基于空间相关性的大规模分布式用户光伏空间分群方法等；（4）面向态势管控的配电网控制、运行、规划及网络安全，典型的如取得突破的配电网不停电转供等。2021年，本项目成果将通过浙电云平台开始大规模推广，另外还将在高弹性主动配电网开展新的探索。