A semantic data framework to support data-driven demand forecasting

This paper presents a prototype semantic data framework for integrating heterogeneous data inputs for data-driven demand forecasting. This framework will be a core feature of a data exchange platform to improve the access and exchange of data between stakeholders involved in the operation and planning of energy systems. Surveys revealed that these stakeholders require reliable data on expected energy production and consumption for strategic and real-time decision-making. A core feature of the framework is the application of semantic technologies for comprehending spatial and temporal data requirements of energy demand forecasting. This paper demonstrates an approach to meeting these semantic requirements through established data standards and models. The conceptual design process followed the following stages: surveying stakeholders, researching digital technologies’ capability, and systematically evaluating the available data. In this paper, we present a prototype based on simulated data. Inputs and results from the simulation model, extracted from open datasets, were structured and stored in a knowledge graph comprised of virtual entities of buildings and geospatial regions. Multiple virtual entities can be linked to a single real-world entity to provide a flexible and adaptable approach to data-driven demand forecasting.

本文提出了一个用于整合异构数据输入以进行数据驱动需求预测的语义数据框架原型。该框架将成为一个数据交换平台的核心特性，以改善能源系统运营和规划所涉及的利益相关者之间的数据获取和交换。调查显示，这些利益相关者在进行战略和实时决策时需要有关预期能源生产和消耗的可靠数据。该框架的一个核心特性是应用语义技术来理解能源需求预测的空间和时间数据要求。本文展示了一种通过已建立的数据标准和模型来满足这些语义要求的方法。概念设计过程遵循以下阶段：对利益相关者进行调查、研究数字技术的能力以及系统地评估可用数据。在本文中，我们基于模拟数据展示了一个原型。从开放数据集中提取的模拟模型的输入和结果经过结构化处理，并存储在一个由建筑物和地理空间区域的虚拟实体组成的知识图谱中。多个虚拟实体可以链接到一个真实世界的实体，从而为数据驱动的需求预测提供一种灵活且适应性强的方法。