Abstract:

In order to detect the spontaneous combustion of coal gangue in open-pit mines in time and avoid greater losses caused by the spread of fire, the technology of spontaneous combustion and fire detection in open-pit mines based on infrared thermal imaging and deep learning was studied. As a result, the problem that a single detection method cannot achieve all-round spontaneous combustion and fire detection of coal gangue in large open-pit mines was solved. First, it was proposed to realize spontaneous combustion and fire detection in different areas through infrared thermal imaging and fire image target recognition, and a spontaneous combustion recognition of coal seam and early warning system architecture of open-pit mines was built. Then, thermal imaging equipment fixed on the ground of the open-pit mine was used to capture infrared images of the coal gangue area where no fire has occurred in real time and monitor the temperature changes of the coal gangue. Finally, the drone with an airborne camera was used to take photos of coal seams, waste piles, and surrounding environments, and then a fire detection model was built through the YOLOv8 target detection algorithm based on deep learning, so as to realize the detection of image targets of flames and smoke, thereby completing fire detection and early warning. The results show that the joint detection technology based on thermal infrared images and flame and smoke image recognition has an average detection accuracy of 70.5% for flames and an average recognition accuracy of 74% for smoke, which can meet the detection needs of spontaneous combustion and fire in open-pit mines.

Key words: infrared thermal imaging, open-pit mine, fire detection, spontaneous combustion of coal seam, deep learning