Abstract
A fire detection research program conducted at the National Institute for Occupational Safety and Health (NIOSH) Pittsburgh Research Laboratory (PRL) recently demonstrated the advantage of multiple fire sensors for early fire detection and nuisance alarm discrimination in underground coal mines. As an example, research has shown that an appropriate combination of smoke, CO, and metal oxide semiconductor (MOS) sensors has the capability to detect a smoldering conveyor belt fire which produces low visibility due to smoke, but CO concentrations too low for an early fire alarm. Such a sensor combination has the additional advantage of being able to distinguish a nuisance alarm event such as those produced by diesel engines or acetylene torches from mine fire products-of-combustion (POC) produced by a real fire. Research has also shown that the problem of hydrogen (H2) gas cross-interference with chemical CO sensors at battery-charging operations in diesel-emissions backgrounds can be resolved with a smoke sensor and a MOS sensor sensitive to NOx associated with diesel emissions. Other underground conditions, such as rock dust, exacerbate the problems with sensors. NIOSH has developed a neural network that takes many of these variables into account as it assesses real-time sensor data to discriminate nuisance alarms.
