IoT-Based Fire Detection: A Comparative Study of Machine Learning Techniques

Abstract

Fires that cannot be detected quickly become uncontrollable. The fires that start to spread uncontrollably pose a significant danger to humans and natural life. Especially in public and crowded areas, fires can lead to possible loss of life and massive property damage. Because of this, it is necessary to detect fires as accurately and quickly as possible. Smoke detectors used with Internet of Things (IoT) technology can exchange data with each other. In this study, data collected from two different types of IoT-based smoke detectors were processed using machine learning algorithms. The k-Nearest Neighbor (k-NN), Multilayer Perceptron (MLP), Radial Basis Function (RBF) Network, Naïve Bayes (NB), Decision Tree (DT), Random Forest (RF), and Logistic Model Tree (LMT) algorithms were used. The data obtained from the smoke detectors were processed using machine learning algorithms to create a highly successful model design. The aim of the study is to design an artificial intelligence-based system that enables the early detection of fires occurring both indoors and outdoors.