ABSTRACT
Underground mine accidents, such as mine fires, remain a concern for mine operators, posing a health and safety risk to the mine workers. Dealing with an unknown location of an accident underground can be a challenging task, creating a hazardous condition for miners during an evacuation and rescue operation. A timely determination of an underground fire event’s location and size is of great importance in reducing the risk of any injuries. Machine learning (ML) has made its way into mining, enabling the development of data-driven predictive models that can be applied to miner’s health and safety problems. A new methodology has been developed using the application of a ML technique to characterize underground accidents such as size and location of an underground mine fire using the post-fire airflow data. This paper describes the methodology and its verification through examples. The National Institute for Occupational Safety and Health (NIOSH) is endeavoring to develop workplace solutions to improve detection of and reduce the risk of hazardous conditions. The results demonstrate a promising application of the ML-based models using the airflow monitoring and provide a useful tool for solving the problem of unknown fire location and reducing the risk of hazardous conditions.
