Unsupervised Learning-Based Behavioural Analysis of IoT-Enabled Air Conditioner Systems Using Environmental Sensing

Abstract

The growing demand for energy-efficient smart indoor environments demands the development of intelligent systems that is capable of real-time monitoring, behavioural analysis, along with autonomous decision-making. This study comprises of the ESP32 based IoT model that supports intelligent monitoring and autonomous decision - making smart indoor system. The environmental data collected using various sensors. This data was logged on to the cloud. To identify the implicit operating behavioural pattern of the data, K-means clustering algorithms along with Elbow method was utilised. For graphical pattern visualization and interpretation, Cluster centroid heatmaps and PCA (Principal Component Analysis) were used. This system enables the transition towards an adaptive and intelligent control system. The smart system is comprised of environmental sensors that results into real time data acquisition and data logging on cloud using Firebase. This research study, helped to monitor the pattern behaviour of the data which is useful for the transition of unsupervised data into supervised data.