输电网知识图谱数据集

数字电网集成了丰富、多样化的资源、设备、功能等，具有高时变性和高复杂性。这种高时变性需要快速、灵活的运行决策技术支持，以保证其安全经济运行，而其高复杂性为快速运行决策带来了严峻挑战。传统解析式的运行规则计算方法需要大量仿真计算，无法保证足够的实时性，数据源单一、精度不高，不能满足精准、智能、快速的智能服务需求。以深度学习为代表的人工智能技术和知识图谱的引入能将数据与知识有机融合——利用知识图谱辅助深度学习，并利用深度学习结果拓展和更新知识图谱，挖掘数据间潜在联系、打破业务壁垒、以及提高各场景中模型的可迁移性，克服传统电力系统输-变-配各业务环节对模型高度依赖、数据价值挖掘与知识利用不足等潜在问题，解决数字电网各环节时变性、复杂性与有限建模能力间的矛盾。基于以上考虑，本节提出融合深度学习和电力知识图谱的数字电网智能服务框架，给出深度学习和电力知识图谱的融合框架和机理，提出了电力知识图谱的构建方法，并通过一个算例分析和展示了融合深度学习和电力知识图谱的解决方案在计算效率、准确性等方面的优越性，为该智能服务框架在解决数字电网兼具安全性和经济性的其他业务场景奠定理论框架基础。数据集包括电力设备实体及属性、规则知识实体及属性、深度学习模型实体及属性、各实体间的关系数据，通过实验试验、计算模拟、融合集成、分析挖掘，基于仿真软件python，pytorch，pandapower,matlab,matpower生成。

The digital power grid integrates abundant and diverse resources, equipment, functions and other elements, and features high time-varying characteristics and complexity. Such high time-varying nature requires fast and flexible operational decision-making technologies to ensure safe and economical operation, while its high complexity poses severe challenges to rapid operational decision-making. Traditional analytical operational rule calculation methods require massive simulation calculations, fail to guarantee sufficient real-time performance, and suffer from single data source and low accuracy, thus unable to meet the demands of precise, intelligent and rapid intelligent services. Introducing artificial intelligence technologies represented by deep learning and knowledge graphs can organically integrate data and knowledge: knowledge graphs are used to assist deep learning, while deep learning results are employed to expand and update knowledge graphs, mine potential connections between data, break through business barriers, and improve model transferability across various scenarios. This approach overcomes potential issues such as the high dependence of traditional power system transmission, transformation and distribution business links on models, insufficient data value mining and knowledge utilization, and resolves the contradiction between the time-varying nature, complexity and limited modeling capabilities of each link in the digital power grid. Based on the above considerations, this section proposes a digital power grid intelligent service framework integrating deep learning and power knowledge graphs, presents the integration framework and mechanism of deep learning and power knowledge graphs, and puts forward the construction method of power knowledge graphs. Additionally, a case study is conducted to analyze and demonstrate the superiority of the solution integrating deep learning and power knowledge graphs in terms of computational efficiency, accuracy and other aspects, laying a theoretical framework foundation for applying this intelligent service framework to other business scenarios of the digital power grid that require both safety and economy. The dataset includes entities and attributes of power equipment, entities and attributes of rule-based knowledge, entities and attributes of deep learning models, as well as relational data between various entities. It is generated through experimental tests, computational simulations, fusion and integration, and analytical mining, based on simulation software including Python, PyTorch, pandapower, MATLAB and MATPOWER.

搜集汇总

背景与挑战

背景概述

该数据集是一个针对数字电网设计的知识图谱数据集，旨在通过融合深度学习和电力知识图谱技术，解决电网高时变性和高复杂性带来的运行决策挑战，提升智能服务的计算效率和准确性。数据集包含电力设备、规则知识、深度学习模型等实体及其属性与关系数据，由清华大学研究人员基于仿真软件生成，数据量为150.64KB，属于国家重点研发计划项目'数字电网关键技术'的成果。