Event Based Energy Disaggregation Algorithm for Autonomous Demand-Ready Response Operations in Homes

In this study, we introduce an event-based unsupervised algorithm focusing on event detection, matching, and clustering for real-time and cost-effective energy disaggregation. The algorithm proposes an iterative matching process for appliance signatures within clusters, reducing errors and misclassifications. It also addresses challenges posed by noisy appliance signatures and determines the optimal number of clusters through a hierarchical clustering scheme, improving disaggregation accuracy. Experimental results from datasets, including UK-DALE and a Turkish household, demonstrate the algorithm's effectiveness, particularly for newly installed or diverse appliances.