A Systematic Energy Information Collection Methodology for Improved Energy Analytics

世界各地的大厦越来越多地配备了传感系统来测量电能消耗，例如智能电表或楼宇自动化系统。它们为能源分析奠定了基础，即从收集的消费数据中提取更高层次的信息。对收集到的数据进行分析，例如通过信号处理或机器学习技术，可以提供预测未来消费模式或将家庭总需求分解为单个家电的贡献等服务。然而，尽管高度依赖信息含量高的消费数据，但迄今尚未出现一种确定如何测量环境以获得适当质量数据的方法。因此，我们将在这个项目中解决这一挑战——寻找一种广泛适用的传感方法，以实现高精度的能源分析。作为确定传感方法的先决条件，我们将开发量化能源数据集合中的信息内容的方法。它们将使我们能够评估基于给定数据的能源分析可以进行到何种程度。随后，我们将推导出一种方法，用于协调部署传感器来收集电能和/或电力数据，以及可能的其他环境参数。我们的方法将主要指定对要部署的感测点的空间和时间分辨率的要求（即所需的采样率和传感器数量），但也将预期的信息增益与感测基础设施的成本联系起来。我们的工作的实际相关性主要是通过在项目本身或其他研究小组中收集的真实场景数据集进行操作来确保的。

Edifices worldwide are increasingly fitted with sensing systems to measure electrical energy consumptions, such as smart meters or building automation systems. They establish the foundation for energy analytics, i.e., the extraction of higher-level information from collected consumption data. The analysis of collected data, e.g., by means of signal processing or machine learning techniques, allows for the provision of services like the prediction of future consumption patterns or the disaggregation of a household total demand into the contributions of individual appliances. Despite the strong reliance on consumption data with a high information content, however, a methodology defining how to instrument the environment in order to attain data of appropriate quality has not emerged to date. We will hence address this challenge - finding a widely applicable sensing methodology to enable energy analytics at high accuracy - within this project. As a prerequisite for the determination of the sensing methodology, we will develop methods to quantify the information content in a collection of energy data. They will allow us to evaluate to which extent energy analytics can be conducted based on the given data. Subsequently, we will derive a methodology for the concerted deployment of sensors to collect electrical energy and/or power data, and possibly also additional environmental parameters. Our methodology will primarily specify requirements to the spatial and temporal resolution of the sensing points to deploy (i.e., required sampling rate and number of sensors), but also relate the expected information gain to the cost of the sensing infrastructure. The practical relevance of our work is ensured by primarily operating on data sets collected in real-world scenarios, either within the project itself or by other research groups.