C12 -- IoT-Enabled Data-Driven Approach: Appliance Clustering and Load Disaggregation for Smart Building Appliance Scheduling ✓ Most Recent

In response to the pressing need for decarbonisation and renewable energy integration, this paper introduces a robust technical solution architecture. This architecture, designed to expedite the shift towards sustainability, optimises electricity usage in buildings. The solution employs a holistic strategy, beginning with the incorporation of IoT (Internet of things) devices, specifi-cally smart plugs, for appliance-level data collection. The architecture comprises two primary engines: the load disaggregation engine, which includes a pattern identification engine and a knapsack optimisation engine, and the clustering engine. The for-mer utilises machine learning algorithms for data preprocessing and optimisation, achieving an average accuracy of 90.2%on unseen data over two days. The latter engine enhances the granularity of appliance behaviour analysis by considering variables such as date, time, season, and weekdays, thereby offering valuable insights into energy consumption patterns. The paper also briefly discusses the day-ahead optimisation engine, underscoring its contribution to energy efficiency and how the introduced engines are going to contribute to the ultimate goal. The architecture, deployed on Microsoft Azure, ensures solu-bility and accessibility, enabling users to access the system via an API (application programming interface), thereby promoting the adoption of energy optimisation practices.