<![CDATA[Fire is one of the most frequent disasters and poses a significant risk to human safety, environmental sustainability, and property due to delayed early detection. This study aims to design and implement an early fire warning system based on the Internet of Things (IoT) enhanced with Machine Learning to improve detection accuracy and reliability. The system utilizes an ESP32 microcontroller as an edge node integrated with a DHT11 sensor for temperature and humidity, an MQ-2 sensor for gas and smoke concentration, and a flame sensor for fire detection. Multisensor data are transmitted in real time to a Flask-based server via the HTTP protocol and processed using a Random Forest classification model to determine environmental conditions as either safe or fire-hazardous. The classification results are displayed on a web-based dashboard and accompanied by automatic notifications delivered through a Telegram bot. Experimental results show that the proposed system achieves a detection accuracy of 94%, a low false positive rate, and a notification latency of less than 3 seconds, based on experiments conducted using a dataset of 3000 samples with an 80:20 split between training and testing data.The integration of IoT and Machine Learning demonstrates superior performance compared to conventional threshold-based methods, making the system a promising preventive solution for fire risk mitigation in residential and industrial environments.]]>
Copyrights © 2026
