A gas detection system combined with a global extension extreme learning machine for early warning of electrical fires

Overheating failures in electrical equipment can lead to fires, making advanced fire warning technologies essential for preventing or limiting the spread of such incidents. In this work, we develop a gas detection system designed to identify volatile gases emitted from various overheated materials in electrical equipment at different heating stages. By integrating the Global Extension Extreme Learning Machine (GEELM), the system effectively classifies gas information related to overheating materials across multiple time intervals. The gas detection system consists of an odor generation equipment and an external high-current generator. Without producing soot particles of smoke, it collects gas information for six types of plastics, including four boards and two cables. Finally, GEELM achieves the optimal performance in classifying gas information from various overheated materials at different heating times. This confirms the system’s feasibility and effectiveness, offering a powerful technical solution for early detection of overheating issues in electrical equipment.