A Smart IoT-Based Fire Detection and Machine Learning Based Control System for Advancing Fire Safety

Abstract

In this study, detailed research has been introduced to enhance the fire safety and security of the chemical laboratories by using the integration of IoT and machine learning. The purpose of the research is to develop a strong framework to identify and take a quick response of any potential fire incidents or safety hazard properly in real time. Using different kinds of sensors, including temperature, smoke, and gas sensors in different locations of the laboratory area, this framework has collected real-time data and identifies the anomalies, which may cause fire or any safety issue. At last, the use of machine learning technologies, including SVM, ANN, DT, and RF helps to understand and analyze the nature of sensor data and help to make a suitable decision for the response. According to the experimental results, the performance of the SVM is excellent in this context, where a precision of 0.987, recall of 0.989, F1 score of 0.988, and AUC-ROC curve of 0.985 has been identified. In addition, the effectiveness of the ANN, DT, and RF is also satisfactory, which can be considered as an effective technology in the context of the fire safety application. By using SVM in the IoT fire detection system, the added advantage has been found in terms of the robustness, interpretability, and computational feasibility, which increases its success ratio. Results of the present study show a significant potential of the IoT and ML, which can be used to redesign the fire safety and emergency response mechanism of the chemical laboratories. Using advanced optimization techniques and sensors, the scalability, efficiency, and reliability of the present framework can be increased.