MQTT Based Power Consumption Monitoring with Usage Pattern Visualization Using Uniform Manifold Approximation and Projection for Smart Buildings

Abstract

The persistent electricity price hikes in the Philippines and the impact of climate change on the country’s energy demands adversely affect the consumer’s finances on a regular basis. These form the foundation for energy efficiency initiatives in building-based electricity consumption. Such initiatives encompass a wide range of innovations from energy generation to real time monitoring, management and control. At its core, this study supplements institutional efforts on energy management through the electricity consumption monitoring system. A non-intrusive data acquisition system that monitors the aggregate electricity consumption and visualizes the appliance usage patterns in a building setting was developed. This Non-Intrusive Load Monitoring (NILM) technique allowed acquisition of data from a single point of measurement using only a single sensor clamped to the main powerline. The acquired data were streamlined to communicate with the IoT OpenHAB framework via Message Queuing Telemetry Transport (MQTT) protocol and were implemented through deployment. Lastly, Uniform Manifold Approximation and Projection (UMAP) was used for dimension reduction. UMAP was applied to the raw time series data of the aggregate power consumption in order to visualize and determine appliance usage patterns, and effectively label data instances through a scatter plot.